diff --git a/CLAUDE.md b/CLAUDE.md index 35db4608f..0ad5a5d00 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -147,7 +147,7 @@ a skill's own bash. **Size budgets:** generated SKILL.md files warn above 160KB (~40K tokens); `test/catalog-budget.test.ts` caps the always-loaded skill catalog at -`CATALOG_BUDGET_TOKEN_EQUIVALENTS` (1,452 today; each new skill ratchets it); and +`CATALOG_BUDGET_TOKEN_EQUIVALENTS` (2,290 today; each new skill ratchets it); and `test/context-budget-ratchet.test.ts` pins per-skill token ceilings against `test/fixtures/context-budget.json` (for legitimate growth or a landed reduction, re-run `bun test/helpers/capture-context-budget.ts` and commit the diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 56f9f0e4e..409090f4b 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -429,7 +429,7 @@ eval files, and misses the strict classifier. No API keys needed. - **Generator tests** (`test/gen-skill-docs.test.ts`) — Tests the template system: verifies placeholders resolve correctly, output includes value hints for flags (e.g. `-d ` not just `-d`), enriched descriptions for key commands (e.g. `is` lists valid states, `press` lists key examples), and that a template's `codex_default_prompt:` frontmatter key reaches the Codex `agents/openai.yaml` `default_prompt` (a template without the key keeps `Use for this task.`) while no host's SKILL.md render carries the key. - **Design detector, catalog, and DESIGN.md** (`test/gstack-design-detect.test.ts`, `test/design-detect-contract.test.ts`, `test/design-catalog.test.ts`, `test/design-checklist-sync.test.ts`, `test/design-md.test.ts`, `test/frontend-scope.test.ts`, `test/impeccable-fixtures.test.ts`) — Drive `bin/gstack-design-detect.ts` through the fake engine in `test/fixtures/fake-impeccable.ts` (probe order, the never-execute-a-repository-file rule, the `--changed` target allow-list, `design_detector: off`, analytics lines, output sanitizing), pin the catalog invariants and the generated `review/design-checklist.md`, round-trip the open DESIGN.md reader/writer, and check the real engine captures (`test/fixtures/impeccable-*.json`, engine 0.1.3) against the contract. `test/dom-dump-hygiene.test.ts` runs `lib/dom-dump.js` in a real Chromium page through the built browse binary; it self-skips without the binary and is opt-in outside CI (`GSTACK_DOM_DUMP_HYGIENE=1`). - **Tier-alignment invariant** (`test/e2e-tier-alignment.test.ts`) — For every self-gated `test/skill-e2e-*.test.ts` named in a touchfiles dep list, the file's `EVALS_TIER` self-gate must match its declared tier in `E2E_TIERS`. Kills the "inert demotion" class where a test is re-tiered in `touchfiles.ts` but the file still gates on the old tier and keeps running in the wrong lane. Unmapped or mixed-tier files are reported, never silently skipped. -- **Catalog budget** (`test/catalog-budget.test.ts`) — Caps the aggregate discovery surface: the sum of every skill's frontmatter `name` + `description` (what every host loads at discovery, every session) must stay under 1,452 token-equivalents, with a 260-byte per-skill cap. Counting goes through the shared census in `test/helpers/skill-census.ts` (physical files vs authored skills vs registry entries — three deliberately different counts). Adding a skill? The failure message carries the re-measure + ratchet protocol. +- **Catalog budget** (`test/catalog-budget.test.ts`) — Caps the aggregate discovery surface: the sum of every skill's frontmatter `name` + `description` (what every host loads at discovery, every session) must stay under 2,290 token-equivalents, with a 260-byte per-skill cap. Counting goes through the shared census in `test/helpers/skill-census.ts` (physical files vs authored skills vs registry entries — three deliberately different counts). Adding a skill? The failure message carries the re-measure + ratchet protocol. - **Context-budget ratchet** (`test/context-budget-ratchet.test.ts`) — CI ceilings on the two token ledgers the catalog budget doesn't cover: the always-on full-frontmatter aggregate and each skill's per-invocation eager tokens (SKILL.md + forced-read references), graded against `test/fixtures/context-budget.json` via `lib/context-bill.ts`. New skills fail until they have a ceiling; ceilings for removed skills must be pruned. Legitimate growth or a landed reduction: re-run `bun test/helpers/capture-context-budget.ts` and commit the refreshed fixture in the same commit, so the change is a visible decision in the diff. - **Skill-to-bin drift ratchet** (`test/skill-bin-drift.test.ts`) — `scripts/skill-bin-drift.ts` statically reads every gstack bin call in skill templates, carved sections, references and the Claude-host renders (it never executes a bin) and fails with `DRIFT_UNKNOWN_BIN` or `DRIFT_UNKNOWN_FLAG` at file:line when a skill names a bin missing from `bin/` or a flag that bin's parser does not accept. Known misses live in `test/fixtures/skill-bin-drift-allowlist.json`, which only shrinks: a new finding fails, and so does an entry that no longer reproduces. Check one skill with `bun run scripts/skill-bin-drift.ts --skill `; bins with permissive parsers or no usage text are reported, never failed. - **Dependency security regressions** (`test/dependency-security.test.ts`) — Run `bun test test/dependency-security.test.ts` to check the resolved `sharp`, `adm-zip`, `qs`, `@hono/node-server` and `hono` version floors, load Sharp, verify ordinary ZIP extraction, and reject extraction through destination-file and destination-directory symlinks. The symlink cases skip Windows. These checks complement the OSV scan; they do not change its existing exceptions. diff --git a/SKILL.md b/SKILL.md index 58ac4edc0..8f2d4aa71 100644 --- a/SKILL.md +++ b/SKILL.md @@ -2,7 +2,7 @@ name: gstack preamble-tier: 1 version: 1.2.0 -description: Router for the gstack skill suite. (gstack) +description: "Picks the gstack skill that fits a request: planning, review, QA, shipping, debugging, docs, security or design. Use for: gstack, which gstack skill, route this with gstack. (gstack)" allowed-tools: - Bash - Read diff --git a/SKILL.md.tmpl b/SKILL.md.tmpl index 3f7cd498e..745251a6e 100644 --- a/SKILL.md.tmpl +++ b/SKILL.md.tmpl @@ -3,8 +3,8 @@ name: gstack preamble-tier: 1 version: 1.2.0 description: | - Router for the gstack skill suite. Sends any gstack request to the right skill - (planning, review, QA, shipping, debugging, docs, security, design). Routes QA by + Picks the gstack skill that fits a request: planning, review, QA, shipping, + debugging, docs, security or design. Routes QA by intent and browser interaction to /browse. Use when you invoke gstack without a specific skill, or ask "which gstack skill fits this?". (gstack) allowed-tools: diff --git a/autoplan/SKILL.md b/autoplan/SKILL.md index 4dba70896..fea3f5ca2 100644 --- a/autoplan/SKILL.md +++ b/autoplan/SKILL.md @@ -2,7 +2,7 @@ name: autoplan preamble-tier: 3 version: 1.0.0 -description: "Auto-review pipeline — runs the CEO, design, eng and DX reviews in order (lean run: one reviewer), auto-deciding. (gstack)" +description: "Auto-review pipeline — runs the CEO, design, eng and DX reviews in order (lean run: one reviewer), auto-deciding. Use for: run all reviews, automatic review pipeline. (gstack)" triggers: - run all reviews - automatic review pipeline diff --git a/benchmark-models/SKILL.md b/benchmark-models/SKILL.md index 54d2533b8..b9c310ee8 100644 --- a/benchmark-models/SKILL.md +++ b/benchmark-models/SKILL.md @@ -2,7 +2,7 @@ name: benchmark-models preamble-tier: 1 version: 1.0.0 -description: Cross-model benchmark for gstack skills. (gstack) +description: "Cross-model benchmark for gstack skills. Use for: cross model benchmark, compare claude gpt gemini. (gstack)" triggers: - cross model benchmark - compare claude gpt gemini diff --git a/benchmark/SKILL.md b/benchmark/SKILL.md index fc6733d8e..38babc394 100644 --- a/benchmark/SKILL.md +++ b/benchmark/SKILL.md @@ -2,7 +2,7 @@ name: benchmark preamble-tier: 1 version: 1.0.0 -description: Web performance regression detection. (gstack) +description: "Web performance regression detection. Use for: web performance benchmark, check page speed. (gstack)" triggers: - web performance benchmark - check page speed diff --git a/browse/SKILL.md b/browse/SKILL.md index 509263edd..970ed2c40 100644 --- a/browse/SKILL.md +++ b/browse/SKILL.md @@ -2,7 +2,7 @@ name: browse preamble-tier: 1 version: 2.0.0 -description: "Drive a real browser through Aside: open a page, read it, click through a flow, take screenshots, check console errors. (gstack)" +description: "Drive a real browser through Aside: open a page, read it, click through a flow, take screenshots, check console errors. Use for: browse a page, open this url, take page screenshot. (gstack)" triggers: - browse a page - open this url diff --git a/canary/SKILL.md b/canary/SKILL.md index a69584705..9bbe4eaa9 100644 --- a/canary/SKILL.md +++ b/canary/SKILL.md @@ -2,7 +2,7 @@ name: canary preamble-tier: 2 version: 1.0.0 -description: Post-deploy canary monitoring. (gstack) +description: "Post-deploy canary monitoring. Use for: monitor after deploy, canary check. (gstack)" allowed-tools: - Bash - Read diff --git a/captain/SKILL.md b/captain/SKILL.md index f5635fa43..e244f53b5 100644 --- a/captain/SKILL.md +++ b/captain/SKILL.md @@ -2,7 +2,7 @@ name: captain preamble-tier: 2 version: 1.0.0 -description: "Run a crew of coding agents from one controller (also /crew): lanes send READY blocks and never merge; the controller merges through trains, keeps the digest, relays approvals, fixes flakes daily. (gstack)" +description: "Run a crew of coding agents from one controller (also /crew): lanes send READY blocks and never merge; the controller merges through trains, keeps the digest, relays approvals, fixes flakes daily. Use for: captain, crew. (gstack)" triggers: - captain - crew diff --git a/careful/SKILL.md b/careful/SKILL.md index bf5aaca88..f91af4062 100644 --- a/careful/SKILL.md +++ b/careful/SKILL.md @@ -1,7 +1,7 @@ --- name: careful version: 0.1.0 -description: Safety guardrails for destructive commands. (gstack) +description: "Safety guardrails for destructive commands. Use for: be careful, warn before destructive, safety mode. (gstack)" triggers: - be careful - warn before destructive diff --git a/changelog.d/claude-skill-descriptions-say-when-to-use.md b/changelog.d/claude-skill-descriptions-say-when-to-use.md new file mode 100644 index 000000000..fdef5b07e --- /dev/null +++ b/changelog.d/claude-skill-descriptions-say-when-to-use.md @@ -0,0 +1 @@ +- **Claude Code's skill list now says when to use each gstack skill, so Opus loads /review and /plan-eng-review when a request fits.** gstack cut every Claude Code description to its first sentence, which says what a skill is but usually not when to load it. In a test with Opus 5.5 in Claude Code, "Review my changes before I land them" loaded /review 0 of 3 times (its line read "Use when you review this PR's diff."), and "plan this feature" loaded no gstack skill in 9 tries. Almost every Claude Code description now keeps a short when-to-use clause after its first sentence. The clause is the skill's own trigger phrases ("Use for: review my changes, look over my diff, review this pr."), or its "Use when" sentence when it has no trigger list; /design-consultation and /eval-plan keep only their first sentence. /review's line no longer assumes a PR exists (local changes and a branch count), /plan-eng-review says it plans a feature before you build it, and the `gstack` router's line says what it does instead of "Router for the gstack skill suite." The skill list Claude Code loads every session grows by about 840 tokens. The clause is added only to Claude Code's trimmed list, so `--catalog-mode=full` and the other hosts don't get it. diff --git a/codex/SKILL.md b/codex/SKILL.md index 782d64d40..35ca68472 100644 --- a/codex/SKILL.md +++ b/codex/SKILL.md @@ -2,7 +2,7 @@ name: codex preamble-tier: 3 version: 1.0.0 -description: OpenAI Codex CLI wrapper — three modes. (gstack) +description: "OpenAI Codex CLI wrapper — three modes. Use for: codex review, second opinion, outside voice challenge. (gstack)" triggers: - codex review - second opinion diff --git a/context-restore/SKILL.md b/context-restore/SKILL.md index 6196ae5a9..e322d4cb5 100644 --- a/context-restore/SKILL.md +++ b/context-restore/SKILL.md @@ -2,7 +2,7 @@ name: context-restore preamble-tier: 2 version: 1.0.0 -description: Restore working context saved earlier by /context-save. (gstack) +description: "Restore working context saved earlier by /context-save. Use for: resume where i left off, restore context, where was i. (gstack)" allowed-tools: - Bash - Read diff --git a/context-save/SKILL.md b/context-save/SKILL.md index 79d87921c..fb9b631fa 100644 --- a/context-save/SKILL.md +++ b/context-save/SKILL.md @@ -2,7 +2,7 @@ name: context-save preamble-tier: 2 version: 1.0.0 -description: Save working context. (gstack) +description: "Save working context. Use for: save progress, save state, save my work, context save. (gstack)" allowed-tools: - Bash - Read diff --git a/cso/SKILL.md b/cso/SKILL.md index ef9ef28f9..7852aecf3 100644 --- a/cso/SKILL.md +++ b/cso/SKILL.md @@ -1,7 +1,7 @@ --- name: cso version: 3.0.0 -description: "Security audit: supported static findings; qualified profiles add reproduction and repair candidates. (gstack)" +description: "Security audit: supported static findings; qualified profiles add reproduction and repair candidates. Use for: security audit, check for vulnerabilities. (gstack)" allowed-tools: - "Bash(~/.claude/skills/gstack/bin/gstack-cso-launcher *)" - "Bash(~/.claude/skills/gstack/bin/gstack-cso-launcher.exe *)" diff --git a/design-html/SKILL.md b/design-html/SKILL.md index b57d25d93..6ea35466d 100644 --- a/design-html/SKILL.md +++ b/design-html/SKILL.md @@ -2,7 +2,7 @@ name: design-html preamble-tier: 2 version: 1.0.0 -description: "Design finalization: generates production-quality Pretext-native HTML/CSS. (gstack)" +description: "Design finalization: generates production-quality Pretext-native HTML/CSS. Use for: build the design, code the mockup, make design real. (gstack)" triggers: - build the design - code the mockup diff --git a/design-motion/SKILL.md b/design-motion/SKILL.md index 2526448e2..46919b3b6 100644 --- a/design-motion/SKILL.md +++ b/design-motion/SKILL.md @@ -2,7 +2,7 @@ name: design-motion preamble-tier: 2 version: 1.0.0 -description: "Design motion: animate an approved design into a seamless MP4 and GIF loop. (gstack)" +description: "Design motion: animate an approved design into a seamless MP4 and GIF loop. Use for: animate this design, make a showcase loop. (gstack)" triggers: - animate this design - make a showcase loop diff --git a/design-review/SKILL.md b/design-review/SKILL.md index 83b1bfadc..a35b90795 100644 --- a/design-review/SKILL.md +++ b/design-review/SKILL.md @@ -2,7 +2,7 @@ name: design-review preamble-tier: 4 version: 2.0.0 -description: "Designer's eye QA: finds visual inconsistency, spacing issues, hierarchy problems, AI slop patterns, and slow interactions — then fixes them; --build checks UI while you build it. (gstack)" +description: "Designer's eye QA: finds visual inconsistency, spacing issues, hierarchy problems, AI slop patterns, and slow interactions — then fixes them; --build checks UI while you build it. Use for: visual design audit, design qa. (gstack)" allowed-tools: - Bash - Read diff --git a/design-shotgun/SKILL.md b/design-shotgun/SKILL.md index 96bfd56b5..1d20e76d2 100644 --- a/design-shotgun/SKILL.md +++ b/design-shotgun/SKILL.md @@ -2,7 +2,7 @@ name: design-shotgun preamble-tier: 2 version: 1.0.0 -description: "Design shotgun: generate multiple AI design variants, open a comparison board, collect structured feedback, and iterate. (gstack)" +description: "Design shotgun: generate multiple AI design variants, open a comparison board, collect structured feedback, and iterate. Use for: explore design variants, show me design options. (gstack)" triggers: - explore design variants - show me design options diff --git a/deslop-shared-libs/SKILL.md b/deslop-shared-libs/SKILL.md index 6c65869de..a59886688 100644 --- a/deslop-shared-libs/SKILL.md +++ b/deslop-shared-libs/SKILL.md @@ -1,7 +1,7 @@ --- name: deslop-shared-libs version: 1.0.0 -description: Find worthwhile shared-code extractions in recent work. (gstack) +description: "Find worthwhile shared-code extractions in recent work. Use for: find code worth sharing. (gstack)" allowed-tools: - Bash - Read diff --git a/devex-review/SKILL.md b/devex-review/SKILL.md index d28cb3f57..2579987b8 100644 --- a/devex-review/SKILL.md +++ b/devex-review/SKILL.md @@ -2,7 +2,7 @@ name: devex-review preamble-tier: 3 version: 1.0.0 -description: Live developer experience audit. (gstack) +description: "Live developer experience audit. Use for: live dx audit, test developer experience. (gstack)" triggers: - live dx audit - test developer experience diff --git a/diagram/SKILL.md b/diagram/SKILL.md index 71bb28a4c..6198b894d 100644 --- a/diagram/SKILL.md +++ b/diagram/SKILL.md @@ -2,7 +2,7 @@ name: diagram preamble-tier: 1 version: 1.0.0 -description: "Turn an English description (or mermaid source) into a diagram triplet: the source, an editable .excalidraw file you can open on excalidraw.com, and rendered SVG + PNG. (gstack)" +description: "Turn an English description (or mermaid source) into a diagram triplet: the source, an editable .excalidraw file you can open on excalidraw.com, and rendered SVG + PNG. Use for: make a diagram, draw a diagram. (gstack)" allowed-tools: - Bash - Read diff --git a/docs/CONTRIBUTOR_REFERENCE.md b/docs/CONTRIBUTOR_REFERENCE.md index 1275bce96..67b601729 100644 --- a/docs/CONTRIBUTOR_REFERENCE.md +++ b/docs/CONTRIBUTOR_REFERENCE.md @@ -20,7 +20,7 @@ cuts to the coverage audit, review army, or voice directive have real quality co A second, harder ceiling guards the DISCOVERY surface: `test/catalog-budget.test.ts` caps the aggregate frontmatter `name` + `description` across all skills at -`CATALOG_BUDGET_TOKEN_EQUIVALENTS` (1,452 today; each new skill ratchets it) +`CATALOG_BUDGET_TOKEN_EQUIVALENTS` (2,290 today; each new skill ratchets it) token-equivalents (260-byte per-skill sub-cap), counted through the shared census in `test/helpers/skill-census.ts`. This one is enforced, not a warning — every host loads the full catalog every session, so growth here taxes every diff --git a/docs/voice-input.md b/docs/voice-input.md index 3401077f6..e45a5867d 100644 --- a/docs/voice-input.md +++ b/docs/voice-input.md @@ -13,7 +13,8 @@ matching surface. Claude Code and Codex choose a skill from its name and description, so a phrase that appears only under `triggers:` does not activate the skill (Codex output drops the field; the GBrain host keeps it). The four core skills (/office-hours, /review, /qa, /ship) carry their `triggers:` phrases in -the description itself: every phrase in the full description, and at least one -in the one-line catalog entry Claude Code reads ("test the site" for /qa, "ship -it" for /ship). The other skills do not, because under the `core` default they -wait for you to type them (docs/proactive.md). +the full description itself. On Claude Code, the one-line catalog entry for most +skills also ends with a short when-to-use clause built from the first few +`triggers:` phrases ("Use for: qa test this, find bugs on site, test the +site."), or from the skill's "Use when" sentence when it has no trigger list; +`--catalog-mode=full` and the other hosts don't get that clause. diff --git a/document-generate/SKILL.md b/document-generate/SKILL.md index 025ab43e2..bb4749ce3 100644 --- a/document-generate/SKILL.md +++ b/document-generate/SKILL.md @@ -2,7 +2,7 @@ name: document-generate preamble-tier: 2 version: 1.0.0 -description: Generate missing documentation from scratch for a feature, module, or entire project. (gstack) +description: "Generate missing documentation from scratch for a feature, module, or entire project. Use for: write docs for this, generate documentation. (gstack)" allowed-tools: - Bash - Read diff --git a/document-release/SKILL.md b/document-release/SKILL.md index 5519f6c25..66d1cf4e0 100644 --- a/document-release/SKILL.md +++ b/document-release/SKILL.md @@ -2,7 +2,7 @@ name: document-release preamble-tier: 2 version: 1.0.0 -description: Release documentation audit. (gstack) +description: "Release documentation audit. Use for: update docs after ship, document what changed. (gstack)" allowed-tools: - Bash - Read diff --git a/freeze/SKILL.md b/freeze/SKILL.md index 39230fbb5..da4b51008 100644 --- a/freeze/SKILL.md +++ b/freeze/SKILL.md @@ -1,7 +1,7 @@ --- name: freeze version: 0.1.0 -description: Restrict file edits to a specific directory for the session. (gstack) +description: "Restrict file edits to a specific directory for the session. Use for: freeze edits to directory, lock editing scope. (gstack)" triggers: - freeze edits to directory - lock editing scope diff --git a/gstack-upgrade/SKILL.md b/gstack-upgrade/SKILL.md index 2a3efc40d..7c351ee1c 100644 --- a/gstack-upgrade/SKILL.md +++ b/gstack-upgrade/SKILL.md @@ -1,7 +1,7 @@ --- name: gstack-upgrade version: 1.1.0 -description: Upgrade gstack to the latest version. +description: "Upgrade gstack to the latest version. Use for: upgrade gstack, update gstack version." triggers: - upgrade gstack - update gstack version @@ -17,8 +17,6 @@ allowed-tools: # /gstack-upgrade -Upgrade gstack to the latest version and show what's new. - ## Inline upgrade flow This section is referenced by all skill preambles when they detect `UPGRADE_AVAILABLE`. diff --git a/gstack-upgrade/SKILL.md.tmpl b/gstack-upgrade/SKILL.md.tmpl index 04a618f99..7b6f24dd4 100644 --- a/gstack-upgrade/SKILL.md.tmpl +++ b/gstack-upgrade/SKILL.md.tmpl @@ -23,8 +23,6 @@ allowed-tools: # /gstack-upgrade -Upgrade gstack to the latest version and show what's new. - ## Inline upgrade flow This section is referenced by all skill preambles when they detect `UPGRADE_AVAILABLE`. diff --git a/gstack/llms.txt b/gstack/llms.txt index 41e78daf0..c85ccfbbb 100644 --- a/gstack/llms.txt +++ b/gstack/llms.txt @@ -33,7 +33,7 @@ Conventions: - [/document-release](document-release/SKILL.md): Release documentation audit. - [/eval-plan](eval-plan/SKILL.md): Preregister a paid eval (bars, arms, decoys, stop rules), dry-run it at zero cost, then run a three-item priced pilot whose spend table gates approval. - [/freeze](freeze/SKILL.md): Restrict file edits to a specific directory for the session. -- [/gstack](SKILL.md): Router for the gstack skill suite. +- [/gstack](SKILL.md): Picks the gstack skill that fits a request: planning, review, QA, shipping, debugging, docs, security or design. - [/gstack-upgrade](gstack-upgrade/SKILL.md): Upgrade gstack to the latest version. - [/guard](guard/SKILL.md): Full safety mode: destructive command warnings + directory-scoped edits. - [/health](health/SKILL.md): Code quality dashboard. @@ -53,12 +53,12 @@ Conventions: - [/plan-ceo-review](plan-ceo-review/SKILL.md): CEO/founder-mode plan review. - [/plan-design-review](plan-design-review/SKILL.md): Designer's eye plan review — interactive, like CEO and Eng review. - [/plan-devex-review](plan-devex-review/SKILL.md): Interactive developer experience plan review. -- [/plan-eng-review](plan-eng-review/SKILL.md): Eng manager-mode plan review. +- [/plan-eng-review](plan-eng-review/SKILL.md): Use when you plan a feature before building it, or review a plan: locks in architecture, data flow, diagrams, edge cases, test coverage and performance. - [/plan-tune](plan-tune/SKILL.md): Self-tuning question sensitivity + developer psychographic for gstack (v1: observational). - [/qa](qa/SKILL.md): Use when you test the site, API or CLI and fix bugs. - [/qa-only](qa-only/SKILL.md): Report browser/API/CLI/job/worker/webhook bugs. - [/retro](retro/SKILL.md): Weekly engineering retrospective. -- [/review](review/SKILL.md): Use when you review this PR's diff. +- [/review](review/SKILL.md): Use when you review a diff before it lands: local changes, a branch or a PR. - [/runaway-stopper](runaway-stopper/SKILL.md): Find runaway Capy threads and subagents, stop or rotate them safely, and prevent repeats. - [/scrape](scrape/SKILL.md): Pull data from a web page through the Aside browser — your real, already signed-in sessions. - [/setup-browser-cookies](setup-browser-cookies/SKILL.md): Import cookies from your real Chromium browser into the headless browse session. diff --git a/guard/SKILL.md b/guard/SKILL.md index ef4b7bd79..58e5b7db2 100644 --- a/guard/SKILL.md +++ b/guard/SKILL.md @@ -1,7 +1,7 @@ --- name: guard version: 0.1.0 -description: "Full safety mode: destructive command warnings + directory-scoped edits. (gstack)" +description: "Full safety mode: destructive command warnings + directory-scoped edits. Use for: full safety mode, guard against mistakes. (gstack)" triggers: - full safety mode - guard against mistakes diff --git a/health/SKILL.md b/health/SKILL.md index 57ac446c2..85925eb4f 100644 --- a/health/SKILL.md +++ b/health/SKILL.md @@ -2,7 +2,7 @@ name: health preamble-tier: 2 version: 1.0.0 -description: Code quality dashboard. (gstack) +description: "Code quality dashboard. Use for: code health check, quality dashboard. (gstack)" triggers: - code health check - quality dashboard diff --git a/ios-clean/SKILL.md b/ios-clean/SKILL.md index 197edd574..c196044c9 100644 --- a/ios-clean/SKILL.md +++ b/ios-clean/SKILL.md @@ -2,7 +2,7 @@ name: ios-clean preamble-tier: 2 version: 1.0.0 -description: "Remove the DebugBridge SPM package and all #if DEBUG wiring from an iOS app. (gstack)" +description: "Remove the DebugBridge SPM package and all #if DEBUG wiring from an iOS app. Use for: clean the ios debug bridge, remove debugbridge. (gstack)" allowed-tools: - Bash - Read diff --git a/ios-design-review/SKILL.md b/ios-design-review/SKILL.md index 87fbd2fe4..3cb826234 100644 --- a/ios-design-review/SKILL.md +++ b/ios-design-review/SKILL.md @@ -2,7 +2,7 @@ name: ios-design-review preamble-tier: 2 version: 1.0.0 -description: Visual design audit for iOS apps on real hardware. (gstack) +description: "Visual design audit for iOS apps on real hardware. Use for: review the ios design, audit the iphone app visuals. (gstack)" allowed-tools: - Bash - Read diff --git a/ios-fix/SKILL.md b/ios-fix/SKILL.md index a9331cba9..8a8af4059 100644 --- a/ios-fix/SKILL.md +++ b/ios-fix/SKILL.md @@ -2,7 +2,7 @@ name: ios-fix preamble-tier: 2 version: 1.0.0 -description: Autonomous iOS bug fixer. (gstack) +description: "Autonomous iOS bug fixer. Use for: fix this ios bug, patch the iphone app. (gstack)" allowed-tools: - Bash - Read diff --git a/ios-qa/SKILL.md b/ios-qa/SKILL.md index 5f6289805..345e542ef 100644 --- a/ios-qa/SKILL.md +++ b/ios-qa/SKILL.md @@ -2,7 +2,7 @@ name: ios-qa preamble-tier: 3 version: 1.0.0 -description: Live-device iOS QA for SwiftUI apps. (gstack) +description: "Live-device iOS QA for SwiftUI apps. Use for: ios qa, test the iphone app, test the ipad app. (gstack)" allowed-tools: - Bash - Read diff --git a/ios-sync/SKILL.md b/ios-sync/SKILL.md index 5e1fd6a27..9fdc0db8d 100644 --- a/ios-sync/SKILL.md +++ b/ios-sync/SKILL.md @@ -2,7 +2,7 @@ name: ios-sync preamble-tier: 2 version: 1.0.0 -description: Regenerate the iOS debug bridge against the latest upstream gstack templates. (gstack) +description: "Regenerate the iOS debug bridge against the latest upstream gstack templates. Use for: resync the ios debug bridge, regenerate ios accessors. (gstack)" allowed-tools: - Bash - Read diff --git a/land-and-deploy/SKILL.md b/land-and-deploy/SKILL.md index d1e5b2f29..8485425aa 100644 --- a/land-and-deploy/SKILL.md +++ b/land-and-deploy/SKILL.md @@ -2,7 +2,7 @@ name: land-and-deploy preamble-tier: 4 version: 1.0.0 -description: Land and deploy workflow. (gstack) +description: "Land and deploy workflow. Use for: merge and deploy, land the pr, ship to production. (gstack)" allowed-tools: - Bash - Read diff --git a/landing-report/SKILL.md b/landing-report/SKILL.md index e302c7ff5..4f5518acd 100644 --- a/landing-report/SKILL.md +++ b/landing-report/SKILL.md @@ -2,7 +2,7 @@ name: landing-report preamble-tier: 2 version: 0.1.0 -description: Read-only queue dashboard for workspace-aware ship. (gstack) +description: "Read-only queue dashboard for workspace-aware ship. Use for: landing report, version queue, ship queue. (gstack)" triggers: - landing report - version queue diff --git a/learn/SKILL.md b/learn/SKILL.md index 96953e0c3..f86482056 100644 --- a/learn/SKILL.md +++ b/learn/SKILL.md @@ -2,7 +2,7 @@ name: learn preamble-tier: 2 version: 1.1.0 -description: Manage project learnings. +description: "Manage project learnings. Use for: show learnings, what have we learned." triggers: - show learnings - what have we learned diff --git a/lib/skill-triggers.ts b/lib/skill-triggers.ts index f0052bc6c..b7d29d9ed 100644 --- a/lib/skill-triggers.ts +++ b/lib/skill-triggers.ts @@ -6,8 +6,9 @@ * - Router lines (lib/harness-register.ts, 160 chars): the description's own * trigger sentence (its first sentence that starts "Use when", "Use to", * "Use for", ...), followed by the description's first sentence when both - * fit. A description with no trigger sentence (the catalog-trimmed Claude - * render) uses the triggers list below instead. + * fit. A description with no trigger sentence, or whose trigger sentence is + * the catalog-trimmed Claude render's short "Use for:" clause, uses the + * triggers list below instead. * - Hermes descriptions (scripts/gen-skill-docs.ts, hosts/hermes.ts): a lead * of at most 57 chars, the part Hermes's skills index shows: "Use for: " * and the skill's `triggers:` list in frontmatter order, as many whole @@ -63,7 +64,8 @@ export function withTriggerLead(description: string, triggers: string[], max: nu /** A router line of at most max chars: the trigger part first, then the description's first sentence when both fit. */ export function routerLine(description: string, triggers: string[], max = 160): string { const first = sentences(description)[0] ?? ''; - const trigger = triggerSentence(description) || triggerList(triggers, max); + const own = triggerSentence(description); + const trigger = (triggers.length && own.startsWith('Use for:') ? '' : own) || triggerList(triggers, max); if (!trigger) return cutWords(first, max); const both = trigger + ' ' + first; return first && first !== trigger && both.length <= max ? both : cutWords(trigger, max); diff --git a/make-pdf/SKILL.md b/make-pdf/SKILL.md index 0a5779c68..75d94ee0c 100644 --- a/make-pdf/SKILL.md +++ b/make-pdf/SKILL.md @@ -2,7 +2,7 @@ name: make-pdf preamble-tier: 1 version: 1.0.0 -description: Turn any markdown file into a publication-quality PDF. (gstack) +description: "Turn any markdown file into a publication-quality PDF. Use for: markdown to pdf, generate pdf, make pdf, export pdf. (gstack)" triggers: - markdown to pdf - generate pdf diff --git a/modernize/SKILL.md b/modernize/SKILL.md index 9d6547299..015ea7f02 100644 --- a/modernize/SKILL.md +++ b/modernize/SKILL.md @@ -2,7 +2,7 @@ name: modernize preamble-tier: 2 version: 0.1.0 -description: "Bring an aging PR onto its current base and decide: MODERNIZE, REWORK or CLOSE. (gstack)" +description: "Bring an aging PR onto its current base and decide: MODERNIZE, REWORK or CLOSE. Use for: modernize this pr, rebase this old pr. (gstack)" triggers: - modernize this pr - rebase this old pr diff --git a/office-hours/SKILL.md b/office-hours/SKILL.md index 81fabee73..aaeb8c89d 100644 --- a/office-hours/SKILL.md +++ b/office-hours/SKILL.md @@ -2,7 +2,7 @@ name: office-hours preamble-tier: 3 version: 2.0.0 -description: YC Office Hours — two modes. (gstack) +description: "YC Office Hours — two modes. Use for: brainstorm this, is this worth building. (gstack)" allowed-tools: - Bash - Read diff --git a/open-gstack-browser/SKILL.md b/open-gstack-browser/SKILL.md index 26a60673b..eb96dcb86 100644 --- a/open-gstack-browser/SKILL.md +++ b/open-gstack-browser/SKILL.md @@ -2,7 +2,7 @@ name: open-gstack-browser preamble-tier: 1 version: 0.2.0 -description: Launch GStack Browser — AI-controlled Chromium with the sidebar extension baked in. +description: "Launch GStack Browser — AI-controlled Chromium with the sidebar extension baked in. Use for: open gstack browser, launch chromium." triggers: - open gstack browser - launch chromium diff --git a/pair-agent/SKILL.md b/pair-agent/SKILL.md index e201678b5..0cd6cba3f 100644 --- a/pair-agent/SKILL.md +++ b/pair-agent/SKILL.md @@ -2,7 +2,7 @@ name: pair-agent preamble-tier: 2 version: 0.1.0 -description: Pair a remote AI agent with your browser. (gstack) +description: "Pair a remote AI agent with your browser. Use for: pair with agent, connect remote agent. (gstack)" triggers: - pair with agent - connect remote agent diff --git a/plan-ceo-review/SKILL.md b/plan-ceo-review/SKILL.md index 6e544cc5c..364c3c693 100644 --- a/plan-ceo-review/SKILL.md +++ b/plan-ceo-review/SKILL.md @@ -2,7 +2,7 @@ name: plan-ceo-review preamble-tier: 3 version: 1.0.0 -description: CEO/founder-mode plan review. (gstack) +description: "CEO/founder-mode plan review. Use for: think bigger, expand scope, strategy review. (gstack)" allowed-tools: - Read - Grep diff --git a/plan-design-review/SKILL.md b/plan-design-review/SKILL.md index 3069dc781..488093936 100644 --- a/plan-design-review/SKILL.md +++ b/plan-design-review/SKILL.md @@ -2,7 +2,7 @@ name: plan-design-review preamble-tier: 3 version: 2.0.0 -description: Designer's eye plan review — interactive, like CEO and Eng review. (gstack) +description: "Designer's eye plan review — interactive, like CEO and Eng review. Use for: design plan review, review ux plan. (gstack)" allowed-tools: - Read - Edit diff --git a/plan-devex-review/SKILL.md b/plan-devex-review/SKILL.md index 3c7b70821..b10d45477 100644 --- a/plan-devex-review/SKILL.md +++ b/plan-devex-review/SKILL.md @@ -2,7 +2,7 @@ name: plan-devex-review preamble-tier: 3 version: 2.0.0 -description: Interactive developer experience plan review. (gstack) +description: "Interactive developer experience plan review. Use for: developer experience review, dx plan review. (gstack)" allowed-tools: - Read - Edit diff --git a/plan-eng-review/SKILL.md b/plan-eng-review/SKILL.md index 6bf41a9d7..b1f1a2a3a 100644 --- a/plan-eng-review/SKILL.md +++ b/plan-eng-review/SKILL.md @@ -2,7 +2,7 @@ name: plan-eng-review preamble-tier: 3 version: 1.0.0 -description: Eng manager-mode plan review. (gstack) +description: "Use when you plan a feature before building it, or review a plan: locks in architecture, data flow, diagrams, edge cases, test coverage and performance. Use for: plan a feature, review architecture, eng plan review. (gstack)" allowed-tools: - Read - Write @@ -12,6 +12,7 @@ allowed-tools: - Bash - WebSearch triggers: + - plan a feature - review architecture - eng plan review - check the implementation plan diff --git a/plan-eng-review/SKILL.md.tmpl b/plan-eng-review/SKILL.md.tmpl index 2ac3b966d..d258ce4d3 100644 --- a/plan-eng-review/SKILL.md.tmpl +++ b/plan-eng-review/SKILL.md.tmpl @@ -4,10 +4,10 @@ preamble-tier: 3 interactive: true version: 1.0.0 description: | - Eng manager-mode plan review. Lock in the execution plan — architecture, - data flow, diagrams, edge cases, test coverage, performance. Walks through - issues interactively with opinionated recommendations. Use when asked to - "review the architecture", "engineering review", or "lock in the plan". (gstack) + Use when you plan a feature before building it, or review a plan: locks in + architecture, data flow, diagrams, edge cases, test coverage and performance. + Use when asked to "review the architecture", "engineering review", or "lock + in the plan". (gstack) voice-triggers: - "tech review" - "technical review" @@ -22,6 +22,7 @@ allowed-tools: - Bash - WebSearch triggers: + - plan a feature - review architecture - eng plan review - check the implementation plan diff --git a/plan-tune/SKILL.md b/plan-tune/SKILL.md index e602bd250..b97f1a4de 100644 --- a/plan-tune/SKILL.md +++ b/plan-tune/SKILL.md @@ -2,7 +2,7 @@ name: plan-tune preamble-tier: 2 version: 1.0.0 -description: "Self-tuning question sensitivity + developer psychographic for gstack (v1: observational). (gstack)" +description: "Self-tuning question sensitivity + developer psychographic for gstack (v1: observational). Use for: tune questions, stop asking me that. (gstack)" triggers: - tune questions - stop asking me that diff --git a/qa-only/SKILL.md b/qa-only/SKILL.md index 7e3382a6a..d75edf967 100644 --- a/qa-only/SKILL.md +++ b/qa-only/SKILL.md @@ -2,7 +2,7 @@ name: qa-only preamble-tier: 4 version: 1.0.0 -description: Report browser/API/CLI/job/worker/webhook bugs. (gstack) +description: "Report browser/API/CLI/job/worker/webhook bugs. Use for: qa report only, just report bugs, test but dont fix. (gstack)" allowed-tools: - Bash - Read diff --git a/qa/SKILL.md b/qa/SKILL.md index 200890297..102e1d571 100644 --- a/qa/SKILL.md +++ b/qa/SKILL.md @@ -2,7 +2,7 @@ name: qa preamble-tier: 4 version: 2.0.0 -description: Use when you test the site, API or CLI and fix bugs. (gstack) +description: "Use when you test the site, API or CLI and fix bugs. Use for: qa test this, find bugs on site, test the site. (gstack)" allowed-tools: - Bash - Read diff --git a/retro/SKILL.md b/retro/SKILL.md index b0c852a36..46308f6d5 100644 --- a/retro/SKILL.md +++ b/retro/SKILL.md @@ -2,7 +2,7 @@ name: retro preamble-tier: 2 version: 2.0.0 -description: Weekly engineering retrospective. (gstack) +description: "Weekly engineering retrospective. Use for: weekly retro, what did we ship. (gstack)" allowed-tools: - Bash - Read diff --git a/review/SKILL.md b/review/SKILL.md index cbe1b1f33..493cfbe2c 100644 --- a/review/SKILL.md +++ b/review/SKILL.md @@ -2,7 +2,7 @@ name: review preamble-tier: 4 version: 1.0.0 -description: Use when you review this PR's diff. (gstack) +description: "Use when you review a diff before it lands: local changes, a branch or a PR. Use for: review my changes, look over my diff, review this pr. (gstack)" allowed-tools: - Bash - Read @@ -14,11 +14,11 @@ allowed-tools: - AskUserQuestion - WebSearch triggers: - - review this pr + - review my changes - look over my diff + - review this pr - code review - check my diff - - review my changes - pre-landing review contract: version: 1 diff --git a/review/SKILL.md.tmpl b/review/SKILL.md.tmpl index 7ae3e4eef..958d49714 100644 --- a/review/SKILL.md.tmpl +++ b/review/SKILL.md.tmpl @@ -3,10 +3,10 @@ name: review preamble-tier: 4 version: 1.0.0 description: | - Use when you review this PR's diff. - Checks it for SQL safety, LLM trust boundaries, conditional side effects and - bugs that pass CI and break production. Fits "look over my diff", "review my - changes", "code review", "check my diff" or "pre-landing review". + Use when you review a diff before it lands: local changes, a branch or a PR. + Finds bugs that pass CI and break production. Fits "review this PR", "look + over my diff", "review my changes", "code review", "check my diff" or + "pre-landing review". Proactively suggest before a merge. (gstack) codex_default_prompt: "Use $gstack-review to review this diff against its base before it lands: find the bugs that pass CI and break in production, fix the clear ones, and list what remains." allowed-tools: @@ -20,11 +20,11 @@ allowed-tools: - AskUserQuestion - WebSearch triggers: - - review this pr + - review my changes - look over my diff + - review this pr - code review - check my diff - - review my changes - pre-landing review contract: version: 1 diff --git a/runaway-stopper/SKILL.md b/runaway-stopper/SKILL.md index 1852de82d..95ebd5a1e 100644 --- a/runaway-stopper/SKILL.md +++ b/runaway-stopper/SKILL.md @@ -2,7 +2,7 @@ name: runaway-stopper preamble-tier: 1 version: 0.1.0 -description: Find runaway Capy threads and subagents, stop or rotate them safely, and prevent repeats. (gstack) +description: "Find runaway Capy threads and subagents, stop or rotate them safely, and prevent repeats. Use for: find runaway threads, stop runaway jobs. (gstack)" triggers: - find runaway threads - stop runaway jobs diff --git a/scrape/SKILL.md b/scrape/SKILL.md index aa1be5480..e0da85af8 100644 --- a/scrape/SKILL.md +++ b/scrape/SKILL.md @@ -2,7 +2,7 @@ name: scrape preamble-tier: 1 version: 2.0.0 -description: Pull data from a web page through the Aside browser — your real, already signed-in sessions. (gstack) +description: "Pull data from a web page through the Aside browser — your real, already signed-in sessions. Use for: scrape this page, get data from, pull from. (gstack)" allowed-tools: - Bash - Read diff --git a/scripts/gen-skill-docs.ts b/scripts/gen-skill-docs.ts index 937ad99d1..f9b368c89 100644 --- a/scripts/gen-skill-docs.ts +++ b/scripts/gen-skill-docs.ts @@ -29,7 +29,7 @@ import type { ModuleFilter } from './resolvers/types'; import { insertRuntimePreludes } from './resolvers/runtime-root'; import { generateSharedPreambleFile, hoistsPreamble, SHARED_PREAMBLE_FILE } from './resolvers/preamble-hoist'; import { extractVoiceTriggers } from './resolvers/voice-triggers'; -import { frontmatterList, withTriggerLead } from '../lib/skill-triggers'; +import { cutWords, frontmatterList, triggerList, triggerSentence, withTriggerLead } from '../lib/skill-triggers'; import { ALL_HOST_NAMES, resolveHostArg, getHostConfig } from '../hosts/index'; import type { HostConfig } from './host-config'; @@ -279,11 +279,14 @@ export { extractVoiceTriggers, processVoiceTriggers }; // asked to..." voice triggers, "Proactively..." routing guidance, and a // "(gstack)" tag. This pile is the always-loaded catalog surface — every // session pays for the full text. The catalog trim splits the description -// into a one-line catalog entry (lead sentence + "(gstack)") that stays in -// the frontmatter and drops the routing/voice prose. The body is not a home -// for it: Claude Code loads a skill body only after the skill is invoked, so -// routing text there can no longer route anything. Voice triggers live on in -// the root router's routing block ({{VOICE_TRIGGER_ROUTING}}). +// into a one-line catalog entry (lead sentence + a short when-to-use clause + +// "(gstack)") that stays in the frontmatter and drops the rest of the +// routing/voice prose. The clause is built by lib/skill-triggers.ts: the +// skill's `triggers:` list ("Use for: a, b."), else its "Use when" sentence. Without it the listing says what a skill is but not when to load +// it, and Opus 5.5 skipped /review on "review my changes" (GSTA-558). The body +// is not a home for routing text: Claude Code loads a skill body only after +// the skill is invoked. Voice triggers live on in the root router's routing +// block ({{VOICE_TRIGGER_ROUTING}}). // // Opt-out: `--catalog-mode=full` keeps v1.44 behavior (no trim, full // description in frontmatter). Use when debugging routing regressions or @@ -362,11 +365,18 @@ export function splitCatalogDescription(description: string): CatalogParts { return { lead, routingProse, voiceLine, hasGstackTag }; } -/** Build the catalog-trimmed `description:` block. */ -export function buildTrimmedDescription(parts: CatalogParts): string { +/** The when-to-use clause gets at most this many chars, and the whole trimmed description at most CATALOG_DESCRIPTION_MAX (test/catalog-budget.test.ts caps name + description at 260 bytes). */ +const CATALOG_WHEN_MAX = 64; +const CATALOG_DESCRIPTION_MAX = 230; + +/** Build the catalog-trimmed `description:` block: the lead, then the when-to-use clause when it fits and adds something. */ +export function buildTrimmedDescription(parts: CatalogParts, triggers: string[] = []): string { const lead = parts.lead.trim(); const suffix = parts.hasGstackTag ? ' (gstack)' : ''; - return `${lead}${suffix}`; + const room = Math.min(CATALOG_WHEN_MAX, CATALOG_DESCRIPTION_MAX - lead.length - suffix.length - 1); + const sentence = triggerSentence(`${lead} ${parts.routingProse}`); + const when = room < 24 ? '' : triggerList(triggers, room) || (sentence && sentence !== lead ? cutWords(sentence, room) : ''); + return when ? `${lead} ${when}${suffix}` : `${lead}${suffix}`; } /** @@ -424,8 +434,8 @@ export function applyCatalogTrim(content: string, skillName: string): { content: const parts = splitCatalogDescription(descText); // If lead + (gstack) is already most of the text, no trim needed. - const trimmedLen = buildTrimmedDescription(parts).length; - if (trimmedLen >= descText.length - 20) return null; + if (buildTrimmedDescription({ ...parts, routingProse: '' }).length >= descText.length - 20) return null; + const newDesc = buildTrimmedDescription(parts, frontmatterList(content, 'triggers')); // Replace description in frontmatter — keep trailing newline so the next // YAML field doesn't collide on the same line as the description value. @@ -433,7 +443,6 @@ export function applyCatalogTrim(content: string, skillName: string): { content: // case: an interior ": " like "Ship workflow: detect..." which a strict YAML // parser reads as a nested mapping and rejects — #1778). toYamlInlineScalar // only quotes when needed, so descriptions without special chars stay plain. - const newDesc = buildTrimmedDescription(parts); // Function replacer (not a string) so a `$` in the description — e.g. a future // skill referencing `$B`/`$D` — can't be interpreted as a `$&`/`$1` replacement // pattern and silently corrupt the frontmatter. diff --git a/setup-browser-cookies/SKILL.md b/setup-browser-cookies/SKILL.md index ae6811673..821cb2664 100644 --- a/setup-browser-cookies/SKILL.md +++ b/setup-browser-cookies/SKILL.md @@ -2,7 +2,7 @@ name: setup-browser-cookies preamble-tier: 1 version: 1.0.0 -description: Import cookies from your real Chromium browser into the headless browse session. (gstack) +description: "Import cookies from your real Chromium browser into the headless browse session. Use for: import browser cookies, login to test site. (gstack)" triggers: - import browser cookies - login to test site diff --git a/setup-deploy/SKILL.md b/setup-deploy/SKILL.md index 9451df76d..8ab73f4c6 100644 --- a/setup-deploy/SKILL.md +++ b/setup-deploy/SKILL.md @@ -2,7 +2,7 @@ name: setup-deploy preamble-tier: 2 version: 1.0.0 -description: Configure deployment settings for /land-and-deploy. +description: "Configure deployment settings for /land-and-deploy. Use for: configure deploy, setup deployment." triggers: - configure deploy - setup deployment diff --git a/setup-gbrain/SKILL.md b/setup-gbrain/SKILL.md index e6b9db5f6..f40480d69 100644 --- a/setup-gbrain/SKILL.md +++ b/setup-gbrain/SKILL.md @@ -2,7 +2,7 @@ name: setup-gbrain preamble-tier: 2 version: 1.0.0 -description: "Set up gbrain for this coding agent: install the CLI, initialize a local PGLite or Supabase brain, register MCP, capture per-remote trust policy. (gstack)" +description: "Set up gbrain for this coding agent: install the CLI, initialize a local PGLite or Supabase brain, register MCP, capture per-remote trust policy. Use for: setup gbrain, install gbrain, connect gbrain. (gstack)" triggers: - setup gbrain - install gbrain diff --git a/ship/SKILL.md b/ship/SKILL.md index 20814e5ab..f074c2fb9 100644 --- a/ship/SKILL.md +++ b/ship/SKILL.md @@ -2,7 +2,7 @@ name: ship preamble-tier: 4 version: 1.0.0 -description: "Ship it: merge the base, run tests, review the diff, bump VERSION and CHANGELOG, push and create a PR. (gstack)" +description: "Ship it: merge the base, run tests, review the diff, bump VERSION and CHANGELOG, push and create a PR. Use for: ship it, create a pr, push to main. (gstack)" allowed-tools: - Bash - Read @@ -17,7 +17,6 @@ triggers: - ship it - create a pr - push to main - - deploy this --- diff --git a/ship/SKILL.md.tmpl b/ship/SKILL.md.tmpl index 36cd86644..a3fb3b996 100644 --- a/ship/SKILL.md.tmpl +++ b/ship/SKILL.md.tmpl @@ -24,7 +24,6 @@ triggers: - ship it - create a pr - push to main - - deploy this --- {{PREAMBLE}} diff --git a/skillify/SKILL.md b/skillify/SKILL.md index 2bc8e3009..bfd4d5390 100644 --- a/skillify/SKILL.md +++ b/skillify/SKILL.md @@ -2,7 +2,7 @@ name: skillify preamble-tier: 2 version: 1.0.0 -description: Codify the most recent successful /scrape flow into a permanent browser-skill on disk. (gstack) +description: "Codify the most recent successful /scrape flow into a permanent browser-skill on disk. Use for: skillify, codify this scrape, save this scrape. (gstack)" allowed-tools: - Bash - Read diff --git a/spec/SKILL.md b/spec/SKILL.md index 55fb27c30..b546227d6 100644 --- a/spec/SKILL.md +++ b/spec/SKILL.md @@ -2,7 +2,7 @@ name: spec preamble-tier: 3 version: 0.1.0 -description: Turn vague intent into a precise, executable spec in five phases. (gstack) +description: "Turn vague intent into a precise, executable spec in five phases. Use for: spec this out, file an issue, write up a ticket. (gstack)" allowed-tools: - Bash - Read diff --git a/sync-gbrain/SKILL.md b/sync-gbrain/SKILL.md index 21e5c3c18..56942cd94 100644 --- a/sync-gbrain/SKILL.md +++ b/sync-gbrain/SKILL.md @@ -2,7 +2,7 @@ name: sync-gbrain preamble-tier: 2 version: 1.0.0 -description: Keep gbrain current with this repo's code and refresh agent search guidance in CLAUDE.md. (gstack) +description: "Keep gbrain current with this repo's code and refresh agent search guidance in CLAUDE.md. Use for: sync gbrain, refresh gbrain, reindex repo. (gstack)" triggers: - sync gbrain - refresh gbrain diff --git a/test-audit/SKILL.md b/test-audit/SKILL.md index 8cd08c460..87f9ed98d 100644 --- a/test-audit/SKILL.md +++ b/test-audit/SKILL.md @@ -2,7 +2,7 @@ name: test-audit preamble-tier: 2 version: 1.0.0 -description: Find low-value or duplicate tests and the test-only code they keep alive. (gstack) +description: "Find low-value or duplicate tests and the test-only code they keep alive. Use for: audit the test suite, find low-value tests. (gstack)" triggers: - audit the test suite - find low-value tests diff --git a/test/catalog-budget.test.ts b/test/catalog-budget.test.ts index 512639ecb..ce2ffbfd5 100644 --- a/test/catalog-budget.test.ts +++ b/test/catalog-budget.test.ts @@ -103,8 +103,20 @@ import { skillCensus } from './helpers/skill-census'; * New-skill ratchet: previous ceiling 1,423 + ceil(115 / 4) = 1,452 * token-equivalents (5,808 bytes), leaving 11 bytes. Existing descriptions * are unchanged. + * ref GST-2774 when-to-use clause on base 0326b14 (2026-11-14) + * result every trimmed Claude description keeps a when-to-use clause of at + * most 64 chars (its triggers list, else its "Use when" sentence; + * scripts/gen-skill-docs.ts buildTrimmedDescription), and /review, + * /plan-eng-review and the root router reworded their leads; + * measured 9,173 bytes = 2,294 token-equivalents, then 9,160 + * bytes = 2,290 after /ship's list dropped "deploy this" (it + * belongs to /land-and-deploy). Deliberate + * growth, not a new skill: in GSTA-558's trials Opus 5.5 skipped + * /review on "review my changes" (0 of 3, 3 of 3 with the full + * text) and loaded no planning skill on "plan this feature" (0 of 9). + * New ceiling: 2,290 token-equivalents (9,160 bytes), leaving 0 bytes. */ -const CATALOG_BUDGET_TOKEN_EQUIVALENTS = 1_452; +const CATALOG_BUDGET_TOKEN_EQUIVALENTS = 2_290; // Largest today: design-consultation at 229 bytes. A description that needs // more than 260 bytes is a body paragraph, not a catalog entry. diff --git a/test/catalog-trim.test.ts b/test/catalog-trim.test.ts index db130d816..49b8d64c2 100644 --- a/test/catalog-trim.test.ts +++ b/test/catalog-trim.test.ts @@ -225,6 +225,37 @@ describe('buildTrimmedDescription', () => { }); expect(out).toBe('Lead with whitespace. (gstack)'); }); + + // GSTA-558: a lead alone says what a skill is, not when to load it, so the Claude catalog keeps a when-to-use clause. + const parts = (lead: string, routingProse = '') => ({ lead, routingProse, voiceLine: null, hasGstackTag: true }); + + test('adds the triggers list as the when-to-use clause', () => { + const out = buildTrimmedDescription(parts('Pre-landing PR review.', 'Use when asked to review.'), ['review my changes', 'look over my diff']); + expect(out).toBe('Pre-landing PR review. Use for: review my changes, look over my diff. (gstack)'); + }); + + test('keeps only the triggers that fit the clause cap', () => { + const out = buildTrimmedDescription(parts('Lead.'), ['first trigger here', 'second trigger here', 'third trigger here', 'fourth trigger here']); + expect(out).toBe('Lead. Use for: first trigger here, second trigger here. (gstack)'); + }); + + test('without triggers, uses the "Use when" sentence from the routing prose', () => { + const out = buildTrimmedDescription(parts('Quick fix.', 'Applies a patch. Use when asked to fix it. Proactively suggest it.')); + expect(out).toBe('Quick fix. Use when asked to fix it. (gstack)'); + }); + + test('a lead that is already the "Use when" sentence is not repeated', () => { + expect(buildTrimmedDescription(parts('Use when you review a diff.', 'Checks it.'))).toBe('Use when you review a diff. (gstack)'); + }); + + test('without triggers or a "Use when" sentence, the lead stands alone', () => { + expect(buildTrimmedDescription(parts('A long lead sentence that says what the skill does.', 'More detail here.'))).toBe('A long lead sentence that says what the skill does. (gstack)'); + }); + + test('a lead that fills the catalog line gets no clause', () => { + const lead = 'x'.repeat(200) + '.'; + expect(buildTrimmedDescription(parts(lead), ['review my changes'])).toBe(lead + ' (gstack)'); + }); }); describe('applyCatalogTrim', () => { @@ -254,7 +285,10 @@ Original body content here. // The body gains no routing section: Claude Code reads a skill body only // after invocation, so routing prose there routes nothing. expect(content).not.toContain('## When to invoke this skill'); - expect(content).not.toContain('Use when asked to do an example task.'); + expect(content.slice(content.indexOf('\n---', 4))).not.toContain('Use when asked to do an example task.'); + // The catalog line keeps the when-to-use sentence (GSTA-558) and drops the rest of the routing prose. + expect(content).toMatch(/^description: "?Example skill:[^\n]*\. Use when asked to do an example task\. \(gstack\)"?\n/m); + expect(content).not.toContain('Proactively suggest'); // Original body still present expect(content).toContain('# Example body'); expect(content).toContain('Original body content here.'); diff --git a/test/fixtures/context-budget.json b/test/fixtures/context-budget.json index f8ca7a814..f952ac2fd 100644 --- a/test/fixtures/context-budget.json +++ b/test/fixtures/context-budget.json @@ -1,74 +1,74 @@ { "_comment": "Context-budget ratchet ceilings (~tokens). Regenerate: bun test/helpers/capture-context-budget.ts. Headroom: alwaysOnTotal x1.05, eagerPerInvocation x1.1. Graded by test/context-budget-ratchet.test.ts via lib/context-bill.ts checkBudget.", - "alwaysOnTotal": 8000, + "alwaysOnTotal": 8647, "eagerPerInvocation": { - "autoplan": 11289, - "benchmark": 8043, - "benchmark-models": 4086, - "browse": 8943, + "autoplan": 11306, + "benchmark": 8420, + "benchmark-models": 4454, + "browse": 9311, "browser-skills/hackernews-frontpage": 371, - "canary": 7738, - "captain": 7211, - "capy/skill/gstack": 5315, - "careful": 1239, - "codex": 9890, - "context-restore": 6287, - "context-save": 5101, - "cso": 4902, + "canary": 7750, + "captain": 7282, + "capy/skill/gstack": 5847, + "careful": 1255, + "codex": 9437, + "context-restore": 6305, + "context-save": 5118, + "cso": 4916, "design-consultation": 10513, - "design-html": 8896, - "design-motion": 4724, - "design-review": 22175, - "design-shotgun": 9318, - "deslop-shared-libs": 3411, - "devex-review": 15303, - "diagram": 4286, - "document-generate": 6013, - "document-release": 4454, - "eval-plan": 5700, - "freeze": 1115, - "gstack": 5350, + "design-html": 8996, + "design-motion": 4738, + "design-review": 22268, + "design-shotgun": 9333, + "deslop-shared-libs": 3420, + "devex-review": 15442, + "diagram": 4529, + "document-generate": 6027, + "document-release": 4469, + "eval-plan": 6271, + "freeze": 1163, + "gstack": 5793, "gstack-upgrade": 6297, - "guard": 1000, - "health": 5816, - "ios-clean": 2436, - "ios-design-review": 2604, - "ios-fix": 2307, - "ios-qa": 5905, - "ios-sync": 2465, - "land-and-deploy": 13585, - "landing-report": 3269, - "learn": 2124, - "make-pdf": 5204, - "modernize": 3772, - "office-hours": 11467, - "open-gstack-browser": 4872, + "guard": 1013, + "health": 5558, + "ios-clean": 2451, + "ios-design-review": 2621, + "ios-fix": 2312, + "ios-qa": 5920, + "ios-sync": 2482, + "land-and-deploy": 8433, + "landing-report": 3283, + "learn": 2198, + "make-pdf": 5493, + "modernize": 3891, + "office-hours": 11480, + "open-gstack-browser": 5118, "openclaw/skills/gstack-openclaw-ceo-review": 2764, "openclaw/skills/gstack-openclaw-office-hours": 4433, "openclaw/skills/gstack-openclaw-retro": 2542, - "pair-agent": 5293, - "plan-ceo-review": 15776, - "plan-design-review": 14687, - "plan-devex-review": 12215, - "plan-eng-review": 11369, - "plan-tune": 9215, - "qa": 7644, - "qa-only": 5726, - "retro": 13484, - "review": 15417, - "runaway-stopper": 4489, - "scrape": 7312, - "setup-browser-cookies": 3507, - "setup-deploy": 5738, - "setup-gbrain": 10009, - "ship": 15913, - "skillify": 5849, - "spec": 9136, - "sync-gbrain": 9749, - "test-audit": 4622, - "ubicloud": 3971, - "unfreeze": 409, - "verify": 2922, - "write-skill": 2264 + "pair-agent": 5375, + "plan-ceo-review": 15849, + "plan-design-review": 14698, + "plan-devex-review": 12229, + "plan-eng-review": 10875, + "plan-tune": 9304, + "qa": 7669, + "qa-only": 5743, + "retro": 13501, + "review": 15723, + "runaway-stopper": 4576, + "scrape": 7559, + "setup-browser-cookies": 3749, + "setup-deploy": 5751, + "setup-gbrain": 10106, + "ship": 16061, + "skillify": 5864, + "spec": 9152, + "sync-gbrain": 9789, + "test-audit": 4636, + "ubicloud": 4059, + "unfreeze": 423, + "verify": 3231, + "write-skill": 2386 } } diff --git a/test/fixtures/golden/claude-ship-SKILL.md b/test/fixtures/golden/claude-ship-SKILL.md index 20814e5ab..f074c2fb9 100644 --- a/test/fixtures/golden/claude-ship-SKILL.md +++ b/test/fixtures/golden/claude-ship-SKILL.md @@ -2,7 +2,7 @@ name: ship preamble-tier: 4 version: 1.0.0 -description: "Ship it: merge the base, run tests, review the diff, bump VERSION and CHANGELOG, push and create a PR. (gstack)" +description: "Ship it: merge the base, run tests, review the diff, bump VERSION and CHANGELOG, push and create a PR. Use for: ship it, create a pr, push to main. (gstack)" allowed-tools: - Bash - Read @@ -17,7 +17,6 @@ triggers: - ship it - create a pr - push to main - - deploy this --- diff --git a/test/fixtures/skill-md-size-allowlist.json b/test/fixtures/skill-md-size-allowlist.json index 1933b8af5..1935d573d 100644 --- a/test/fixtures/skill-md-size-allowlist.json +++ b/test/fixtures/skill-md-size-allowlist.json @@ -5,53 +5,53 @@ "agy/gstack-autoplan": 109449, "agy/gstack-plan-design-review": 109103, "agy/gstack-plan-devex-review": 114944, - "agy/gstack-plan-eng-review": 143580, - "agy/gstack-review": 104735, + "agy/gstack-plan-eng-review": 143530, + "agy/gstack-review": 104723, "codex/gstack-autoplan": 108192, "codex/gstack-plan-design-review": 110460, "codex/gstack-plan-devex-review": 116452, - "codex/gstack-plan-eng-review": 144745, - "codex/gstack-review": 104119, + "codex/gstack-plan-eng-review": 144695, + "codex/gstack-review": 104107, "copilot/gstack-autoplan": 108892, "copilot/gstack-plan-design-review": 109369, "copilot/gstack-plan-devex-review": 114406, - "copilot/gstack-plan-eng-review": 143745, - "copilot/gstack-review": 104316, + "copilot/gstack-plan-eng-review": 143695, + "copilot/gstack-review": 104304, "cursor/gstack-autoplan": 108124, "cursor/gstack-plan-design-review": 107911, "cursor/gstack-plan-devex-review": 113676, - "cursor/gstack-plan-eng-review": 142335, - "cursor/gstack-review": 122788, + "cursor/gstack-plan-eng-review": 142285, + "cursor/gstack-review": 122776, "factory/gstack-autoplan": 110402, "factory/gstack-plan-design-review": 111009, "factory/gstack-plan-devex-review": 117039, - "factory/gstack-plan-eng-review": 145443, - "factory/gstack-review": 126561, + "factory/gstack-plan-eng-review": 145393, + "factory/gstack-review": 126549, "gbrain/gstack-autoplan": 108531, - "gbrain/gstack-plan-eng-review": 124963, + "gbrain/gstack-plan-eng-review": 124932, "gemini/gstack-autoplan": 108682, "gemini/gstack-plan-design-review": 109197, "gemini/gstack-plan-devex-review": 114234, - "gemini/gstack-plan-eng-review": 143578, - "gemini/gstack-review": 104138, + "gemini/gstack-plan-eng-review": 143528, + "gemini/gstack-review": 104126, "hermes/gstack-autoplan": 108536, - "hermes/gstack-plan-eng-review": 124999, + "hermes/gstack-plan-eng-review": 124885, "kiro/gstack-autoplan": 107898, "kiro/gstack-plan-design-review": 107757, "kiro/gstack-plan-devex-review": 113528, - "kiro/gstack-plan-eng-review": 142197, - "kiro/gstack-review": 122612, + "kiro/gstack-plan-eng-review": 142147, + "kiro/gstack-review": 122600, "openclaw/gstack-autoplan": 108354, - "openclaw/gstack-plan-eng-review": 123478, + "openclaw/gstack-plan-eng-review": 123428, "opencode/gstack-autoplan": 108873, "opencode/gstack-plan-design-review": 109225, "opencode/gstack-plan-devex-review": 114284, - "opencode/gstack-plan-eng-review": 143604, - "opencode/gstack-review": 123410, + "opencode/gstack-plan-eng-review": 143554, + "opencode/gstack-review": 123398, "slate/gstack-autoplan": 108011, "slate/gstack-plan-design-review": 107834, "slate/gstack-plan-devex-review": 113602, - "slate/gstack-plan-eng-review": 142266, - "slate/gstack-review": 122700 + "slate/gstack-plan-eng-review": 142216, + "slate/gstack-review": 122688 } } diff --git a/test/helpers/carve-guards.ts b/test/helpers/carve-guards.ts index d45ec71e1..9e4940ac5 100644 --- a/test/helpers/carve-guards.ts +++ b/test/helpers/carve-guards.ts @@ -161,7 +161,7 @@ export const CARVE_GUARDS: Record = { }, behavioral: 'external', externalTest: 'test/skill-e2e-ship-section-loading.test.ts', - maxSkeletonBytes: 61_438, // GST-2726 (2026-11-15, Garry: bring /plan-eng-review back): Step 1's DIFF hint and the no-Eng-Review line name /plan-eng-review again, the wording GST-1698 removed; measured 61,438 (before: 61,417 on 81fb900). Before that: GSTA-152 token cap (2026-10-29, Garry's "Do it" on plan v3): the skeleton's Budget paragraph (the --token-cap, --resume and --continue flags, start/check/close, the budget-handoff pointer) and the section index's budget-handoff line; measured 61,420 (before: 60,958 on edf4c2e). Before that: GSTA-131 token efficiency (2026-10-27, Garry's D1): Step 1's delta carry (dry run, pregate at HEAD, then carry --pregate) and Step 11.5's fix-check and planned-skip binding, +60 bytes after tightening Step 1 and the Review policy paragraph; measured 60,960 (before: 60,900 on 10b77de4). Before that: GSTA-96 hoist merged with GSTA-92 phase 2 (2026-10-26): the hoisted skeleton (59,706) plus this PR's growth, keeping the hoist's 65-byte headroom; measured 60,848. Before: + v2.0 {{ASIDE_RESEARCH}} (Aside first, WebSearch fallback); measured 78_275. + gate integrity (HONEST_WORK_GATE_INTEGRITY W3): Step 16 stage 4 ZERO-RUN/unknown receipt rows, stage 5 honesty inventory, Step 17 open-gate ready-PR gate; measured 80,540 (2026-10-10). + rule 22 / GBRA-108 (2026-10-25): Step 22's terminal line (gstack-ship-receipt result, the GSTACK_RESULT line a parent greps) and Step 18's unattended routing (an open PR at HEAD gets its body updated, else handoff), offset in part by tightening Steps 14, 16, 18 and 20 prose after the 2026-10-14 tightening left 2 bytes of headroom; measured 81,535. + harness lane E (2026-10-26, issue row 3: four Captains re-ran /ship on identical product code): Step 1's carry check (gstack-ship-receipt carry binds an earlier receipt to HEAD when the product patch-id is unchanged); measured 82,165. + the Review policy paragraph (the blocking floor, pass structure and cap, NO FINDINGS, the review_loop receipt counters, the review_scope knob), stated in the skeleton because gstack is the canonical holder of its review policy (docs/review-policy.md); measured 82,645. + maxSkeletonBytes: 61_467, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is), and /ship's list drops "deploy this", which belongs to /land-and-deploy; measured 61,467 (before: 61,438 on a141e94). Before that: GST-2726 (2026-11-15, Garry: bring /plan-eng-review back): Step 1's DIFF hint and the no-Eng-Review line name /plan-eng-review again, the wording GST-1698 removed; measured 61,438 (before: 61,417 on 81fb900). Before that: GSTA-152 token cap (2026-10-29, Garry's "Do it" on plan v3): the skeleton's Budget paragraph (the --token-cap, --resume and --continue flags, start/check/close, the budget-handoff pointer) and the section index's budget-handoff line; measured 61,420 (before: 60,958 on edf4c2e). Before that: GSTA-131 token efficiency (2026-10-27, Garry's D1): Step 1's delta carry (dry run, pregate at HEAD, then carry --pregate) and Step 11.5's fix-check and planned-skip binding, +60 bytes after tightening Step 1 and the Review policy paragraph; measured 60,960 (before: 60,900 on 10b77de4). Before that: GSTA-96 hoist merged with GSTA-92 phase 2 (2026-10-26): the hoisted skeleton (59,706) plus this PR's growth, keeping the hoist's 65-byte headroom; measured 60,848. Before: + v2.0 {{ASIDE_RESEARCH}} (Aside first, WebSearch fallback); measured 78_275. + gate integrity (HONEST_WORK_GATE_INTEGRITY W3): Step 16 stage 4 ZERO-RUN/unknown receipt rows, stage 5 honesty inventory, Step 17 open-gate ready-PR gate; measured 80,540 (2026-10-10). + rule 22 / GBRA-108 (2026-10-25): Step 22's terminal line (gstack-ship-receipt result, the GSTACK_RESULT line a parent greps) and Step 18's unattended routing (an open PR at HEAD gets its body updated, else handoff), offset in part by tightening Steps 14, 16, 18 and 20 prose after the 2026-10-14 tightening left 2 bytes of headroom; measured 81,535. + harness lane E (2026-10-26, issue row 3: four Captains re-ran /ship on identical product code): Step 1's carry check (gstack-ship-receipt carry binds an earlier receipt to HEAD when the product patch-id is unchanged); measured 82,165. + the Review policy paragraph (the blocking floor, pass structure and cap, NO FINDINGS, the review_loop receipt counters, the review_scope knob), stated in the skeleton because gstack is the canonical holder of its review policy (docs/review-policy.md); measured 82,645. minUnionBytes: 181_000, // token-reduction Phases 1-2 (v1.69.x branch); measured union 201,464 mustContain: ['VERSION', 'CHANGELOG', 'review', 'merge', 'PR'], // v1.58.5.0: pre-push-guard install (#2077) stacks on the shared first-run-guidance preamble. @@ -251,7 +251,7 @@ export const CARVE_GUARDS: Record = { // tier-2+ skeleton (measured 89,184). Main's v1.64.0.0 adds ~340 B more // (telemetry --error-message/--failed-step preamble prose, PR #769). // Budget covers the sum of both waves. - maxSkeletonBytes: 56_209, // GSTA-134 skill diet (2026-10-28): lowered from 66_644 to the measured dieted skeleton, 56,209. Before: GSTA-96 hoist (2026-10-26): lowered from 85_900 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 581 bytes; measured 66,063. Before: Harness-aware outside voice: validated dispatch and provenance. + v1.91.19.0 wave: design --version probe (B4) and verdict-form outside voice (INV-1) on top of v1.91.17.0's round accounting (#3022); measured 80,443 (2026-10-03). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); skeleton measured 84,669 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; skeleton measured 84,804 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; skeleton measured 84,863 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; skeleton measured 84,748 (2026-10-07). Merged with the guard area: measured skeleton 84,945 (2026-10-07). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + maxSkeletonBytes: 56_252, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 56,252 (before: 56,209 on a141e94). Before that: GSTA-134 skill diet (2026-10-28): lowered from 66_644 to the measured dieted skeleton, 56,209. Before: GSTA-96 hoist (2026-10-26): lowered from 85_900 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 581 bytes; measured 66,063. Before: Harness-aware outside voice: validated dispatch and provenance. + v1.91.19.0 wave: design --version probe (B4) and verdict-form outside voice (INV-1) on top of v1.91.17.0's round accounting (#3022); measured 80,443 (2026-10-03). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); skeleton measured 84,669 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; skeleton measured 84,804 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; skeleton measured 84,863 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; skeleton measured 84,748 (2026-10-07). Merged with the guard area: measured skeleton 84,945 (2026-10-07). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). minUnionBytes: 99_200, // token-reduction Phases 1-2 (v1.69.x branch); measured union 110,293 mustContain: ['design', 'visual'], maxSizeRatio: 1.12, // D1 1.104 + main's ~0.008 @@ -356,10 +356,10 @@ do not launch the downstream skill or open a browser.`, // the #538 opt-out + D1 evidence directive — ratio 1.104 measured. // #2499 project-scope MCP jq in the brain-sync block grew every tier-2+ // skeleton ~1.5KB (entry resolution emitted once per SKILL.md). - maxSkeletonBytes: 43_885, // GSTA-134 skill diet (2026-10-28): lowered from 80_533 to the measured skeleton after the second opinion and visual design moved to sections, 43,885. Before: GSTA-96 hoist (2026-10-26): lowered from 95_900 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 952 bytes; measured 79,581. Before: Office-hours + sketch outside voices include host guards and completion checks. + v1.91.19.0 wave: Q open-question prose form (#2729, approved Q3), DESIGN_READY --version probe (B4), Aside ~/.local/bin fallback (E7), verdict-form outside voice (INV-1), plus v1.91.17.0's round accounting (#3022) merged in; measured 89,684 (2026-10-03). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); skeleton measured 92,590 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; skeleton measured 92,869 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; skeleton measured 92,930 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; skeleton measured 92,756 (2026-10-07). Merged with the guard area: measured skeleton 93,099 (2026-10-07). + fix wave PR D: the Phase 4.5 builder-profile read before the session log (#2801) and the pre-telemetry design-doc check (#1049); measured skeleton 94,877 (2026-10-07). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + maxSkeletonBytes: 43_937, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 43,937 (before: 43,885 on a141e94). Before that: GSTA-134 skill diet (2026-10-28): lowered from 80_533 to the measured skeleton after the second opinion and visual design moved to sections, 43,885. Before: GSTA-96 hoist (2026-10-26): lowered from 95_900 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 952 bytes; measured 79,581. Before: Office-hours + sketch outside voices include host guards and completion checks. + v1.91.19.0 wave: Q open-question prose form (#2729, approved Q3), DESIGN_READY --version probe (B4), Aside ~/.local/bin fallback (E7), verdict-form outside voice (INV-1), plus v1.91.17.0's round accounting (#3022) merged in; measured 89,684 (2026-10-03). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); skeleton measured 92,590 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; skeleton measured 92,869 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; skeleton measured 92,930 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; skeleton measured 92,756 (2026-10-07). Merged with the guard area: measured skeleton 93,099 (2026-10-07). + fix wave PR D: the Phase 4.5 builder-profile read before the session log (#2801) and the pre-telemetry design-doc check (#1049); measured skeleton 94,877 (2026-10-07). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). minUnionBytes: 115_800, // Phase 4 wave 4; measured union 118,175 mustContain: ['design doc', 'problem statement'], - maxSizeRatio: 1.1990, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1990 (before: 1.1968). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1968 (before: 1.1922). Before that: GST-2726 (2026-11-15, Garry: bring /plan-eng-review back): the handoff menu, its default and the design-doc line name /plan-eng-review again, +42 B union; measured 156,298 / 131,110 = 1.19211, rounded up (before: 156,256 = 1.19179 on 81fb900). Before that: GSTA-134 (2026-10-28, raised from 1.1796 by GStack Minder as merge-queue owner): +1,662 B union: second-opinion question moved into the main file while CODEX_SECOND_OPINION keeps its copy; per-run load 73,248→43,885 B; measured 156,286 / 131,110 = 1.19202, rounded up. Before: + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); measured 149,758 / 131,110 = 1.1422 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; measured 150,037 / 131,110 = 1.1444 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; measured 150,098 / 131,110 = 1.1448 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; measured 150,011 / 131,110 = 1.1442 (2026-10-07). Merged with the guard area: measured 150,354 / 131,110 = 1.1468 (2026-10-07). + fix wave PR D: builder-profile read before the session log with Phase 6 using the carried values (#2801), the inline design doc before approval (#879) and the pre-telemetry design-doc check (#1049); measured 153,206 / 131,110 = 1.1685 (2026-10-07). + empty LAST_ASSIGNMENT/LAST_PROJECT guards in the welcome_back and regular greetings (backfilled profiles); measured 153,469 / 131,110 = 1.1705 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file and its own copies of shared sections; measured 160,989 / 131,110 = 1.2279 (2026-10-25). Before: 1.18. - GSTA-96 dedupe: the Q open-question form is office-hours' own appended subsection, so its copies of the shared Format and fallback subsections go; measured 154,653 / 131,110 = 1.1796 (2026-10-26), 154,624 = 1.1793 after the main merge. Before: 1.2279. + maxSizeRatio: 1.1993, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 157,236 / 131,110 = 1.19927, rounded up (before: 1.1990). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1990 (before: 1.1968). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1968 (before: 1.1922). Before that: GST-2726 (2026-11-15, Garry: bring /plan-eng-review back): the handoff menu, its default and the design-doc line name /plan-eng-review again, +42 B union; measured 156,298 / 131,110 = 1.19211, rounded up (before: 156,256 = 1.19179 on 81fb900). Before that: GSTA-134 (2026-10-28, raised from 1.1796 by GStack Minder as merge-queue owner): +1,662 B union: second-opinion question moved into the main file while CODEX_SECOND_OPINION keeps its copy; per-run load 73,248→43,885 B; measured 156,286 / 131,110 = 1.19202, rounded up. Before: + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); measured 149,758 / 131,110 = 1.1422 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; measured 150,037 / 131,110 = 1.1444 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; measured 150,098 / 131,110 = 1.1448 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; measured 150,011 / 131,110 = 1.1442 (2026-10-07). Merged with the guard area: measured 150,354 / 131,110 = 1.1468 (2026-10-07). + fix wave PR D: builder-profile read before the session log with Phase 6 using the carried values (#2801), the inline design doc before approval (#879) and the pre-telemetry design-doc check (#1049); measured 153,206 / 131,110 = 1.1685 (2026-10-07). + empty LAST_ASSIGNMENT/LAST_PROJECT guards in the welcome_back and regular greetings (backfilled profiles); measured 153,469 / 131,110 = 1.1705 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file and its own copies of shared sections; measured 160,989 / 131,110 = 1.2279 (2026-10-25). Before: 1.18. - GSTA-96 dedupe: the Q open-question form is office-hours' own appended subsection, so its copies of the shared Format and fallback subsections go; measured 154,653 / 131,110 = 1.1796 (2026-10-26), 154,624 = 1.1793 after the main merge. Before: 1.2279. }, 'document-release': { skill: 'document-release', @@ -490,10 +490,10 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, // operational multi-STOP skill, like ship }, behavioral: 'plan', - maxSkeletonBytes: 60_070, // GST-1088 lane B follow-up (raised from 60,014 to measured, no headroom): the Done-means check/waive commands run through the host-resolved bin path (~/.claude/skills/gstack/bin/gstack-skill-contract) instead of a bare gstack-skill-contract that a standard install doesn't put on PATH (fix-check finding on #3909 at 58aeee7), +56 bytes; measured 60,070. Before: GST-1088 lane B (#3909, raised from 59,774 to measured, no headroom; owner ruling via GStack Minder 2026-11-03): CONTRACT_CHECKER_SHIPPED=true now that bin/gstack-skill-contract ships (#3912) renders the gstack-skill-contract check/waive sentence under the Done means list (renders/GSTA-256/plan-v3.md line 74, r2-d6 "Done-means list"), +240 bytes; lane C's diet is expected to bring /review back under 59,073; measured 60,014. Before: GST-1088 lane B (raised from 59,073 to measured, no headroom): the contract: frontmatter block and the F1 "Done means" list rendered from it at the top of the body, plus widened triggers; F1 is a gate-rev-2 recommendation on the plan Garry approved with "do it" on gate rev 1 (2026-11-03), contingent on his gate-rev-2 answer, and lane C's diet is expected to bring /review back under 59,073; measured 59,774. Before: GSTA-96 hoist (2026-10-26): lowered from 80_900 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 319 bytes; measured 58,754. Before: Shared-code identity/skip/action rules + critical-severity validation; measured 74,493 (2026-09-17). + v1.91.12.0 merge of #2999 (review clarity repairs: await reads, research alongside dispatch, /review deadline and setup authority, findings sources) with #3002 (guarded state-root lines, plan-check checkpoints); each fit alone; measured 74,881 (2026-10-01). + Oct 8 wave: shared voice directive bans "load-bearing" and adds the reply-language rule, offset by tightening the bounded closer (net +17 bytes per tier-2+ skill); measured 74,884 (2026-10-08). + gate integrity (HONEST_WORK_GATE_INTEGRITY W2): Step 3.5 Diff scans with the shared GATE_SCAN_BLOCK, the Step 4 carry, the four-option `review-gate-disposition` question in 5c, gate record fields in 5.8 and the Gate edits report line; measured 80,054 (2026-10-10). + Oct 10 multi-agent wave PR D2 (sideways review): Step 4's enum-only paragraph becomes the sideways sweep with the forced-CRITICAL list, plus the flake-fix number sentence in the Gate Integrity paragraph; measured skeleton 80,812 (2026-10-10). + maxSkeletonBytes: 60_176, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 60,176 (before: 60,070 on a141e94). Before that: GST-1088 lane B follow-up (raised from 60,014 to measured, no headroom): the Done-means check/waive commands run through the host-resolved bin path (~/.claude/skills/gstack/bin/gstack-skill-contract) instead of a bare gstack-skill-contract that a standard install doesn't put on PATH (fix-check finding on #3909 at 58aeee7), +56 bytes; measured 60,070. Before: GST-1088 lane B (#3909, raised from 59,774 to measured, no headroom; owner ruling via GStack Minder 2026-11-03): CONTRACT_CHECKER_SHIPPED=true now that bin/gstack-skill-contract ships (#3912) renders the gstack-skill-contract check/waive sentence under the Done means list (renders/GSTA-256/plan-v3.md line 74, r2-d6 "Done-means list"), +240 bytes; lane C's diet is expected to bring /review back under 59,073; measured 60,014. Before: GST-1088 lane B (raised from 59,073 to measured, no headroom): the contract: frontmatter block and the F1 "Done means" list rendered from it at the top of the body, plus widened triggers; F1 is a gate-rev-2 recommendation on the plan Garry approved with "do it" on gate rev 1 (2026-11-03), contingent on his gate-rev-2 answer, and lane C's diet is expected to bring /review back under 59,073; measured 59,774. Before: GSTA-96 hoist (2026-10-26): lowered from 80_900 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 319 bytes; measured 58,754. Before: Shared-code identity/skip/action rules + critical-severity validation; measured 74,493 (2026-09-17). + v1.91.12.0 merge of #2999 (review clarity repairs: await reads, research alongside dispatch, /review deadline and setup authority, findings sources) with #3002 (guarded state-root lines, plan-check checkpoints); each fit alone; measured 74,881 (2026-10-01). + Oct 8 wave: shared voice directive bans "load-bearing" and adds the reply-language rule, offset by tightening the bounded closer (net +17 bytes per tier-2+ skill); measured 74,884 (2026-10-08). + gate integrity (HONEST_WORK_GATE_INTEGRITY W2): Step 3.5 Diff scans with the shared GATE_SCAN_BLOCK, the Step 4 carry, the four-option `review-gate-disposition` question in 5c, gate record fields in 5.8 and the Gate edits report line; measured 80,054 (2026-10-10). + Oct 10 multi-agent wave PR D2 (sideways review): Step 4's enum-only paragraph becomes the sideways sweep with the forced-CRITICAL list, plus the flake-fix number sentence in the Gate Integrity paragraph; measured skeleton 80,812 (2026-10-10). minUnionBytes: 89_000, // Phase 4 wave 1; measured union 93,357 mustContain: ['confidence', 'P1', 'P2', 'Review Army', 'adversarial'], - maxSizeRatio: 1.3369, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.3369 (before: 1.3343). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.3343 (before: 1.3288). Before that: + GST-1088 lane B follow-up (raised from 1.3283 to measured, no headroom): the Done-means check/waive commands run through the host-resolved bin path instead of a bare gstack-skill-contract that isn't on PATH (fix-check finding on #3909 at 58aeee7), +56 bytes; measured 144,203 / 108,523 = 1.328778, rounded up. Before: + GST-1088 lane B (#3909, raised from 1.3261 to measured, no headroom; owner ruling via GStack Minder 2026-11-03): the gstack-skill-contract check/waive sentence that CONTRACT_CHECKER_SHIPPED=true renders (renders/GSTA-256/plan-v3.md line 74, r2-d6 "Done-means list"), +240 bytes; lane C's diet is expected to bring /review back under 1.3180; measured 144,147 / 108,523 = 1.32826, rounded up. Before: + GST-1088 lane B (raised from 1.3180 to measured, no headroom): the contract: frontmatter block and the F1 "Done means" list, contingent on Garry's gate-rev-2 answer; lane C's diet is expected to bring /review back under 1.3180; measured 143,907 / 108,523 = 1.3261. Before: + GSTA-167 variant G part 1 (plan approved by Garry 2026-10-30, autoplan-20261030161754 "Do it", which names about +0.3k prompt tokens per voice): the native adversarial and outside-voice prompts carry the blocking floor and the Severity and Blocks lines, +537 bytes; measured 1.3180 (2026-10-30). Before: 1.3136. GSTA-96 hoist merged with GSTA-92 phase 2 (2026-10-26): both owner-approved raises together, measured 142,550 / 108,523 = 1.3135. Before: Shared-code feature + critical-severity validation: 128,042 union bytes / 108,523 baseline = 1.1799; preserves content floors. + v1.91.12.0 merge of #2999 (above, plus plan-completion fallback intent and specialist checklist-by-path) with #3002; measured 1.1843 (2026-10-01). + v1.91.19.0 wave: verdict-form Codex callers with sandbox preflight (INV-1/B1), P0/P1 structured gate (B1b), bound-plan discovery and data-not-instructions guard (F1/D3), visible LEARNINGS failures (B5); measured 129,941 / 108,523 = 1.1974 (2026-10-03). + v1.91.18.0 (#3023) merged into the v1.91.19.0 wave; measured 130,025 / 108,523 = 1.1981 (2026-10-04). + Oct 6 fix wave: quota_exhausted preflight branch (#3051); measured 130,359 / 108,523 = 1.2012 (2026-10-06). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); measured 131,314 / 108,523 = 1.2100 (2026-10-06). Combined Oct 6 wave head (A and B): measured 131,966 / 108,523 = 1.2160 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; measured 132,199 / 108,523 = 1.2182 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; measured 132,258 / 108,523 = 1.2187 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; measured 132,233 / 108,523 = 1.2185 (2026-10-07). Merged with the guard area: measured 132,522 / 108,523 = 1.2211 (2026-10-07). + Oct 8 wave: the defensive-security framing and fixture summary mode reach the security specialist and the Red Team (stated once in the dispatch, referenced by the Red Team); measured 133,902 / 108,523 = 1.2339 (2026-10-08). + gate integrity W2 (above); measured 1.282 (2026-10-10). + Oct 10 multi-agent wave PR D2 (sideways review): sideways sweep and flake-fix number in Step 4; `Diff read` and `Undeclared behavior changes` in the review-mode Scope Check (plan-completion section); measured 140,497 / 108,523 = 1.2946 (2026-10-10). + harness lane E (2026-10-26): the lead-judgment stage in the finding merge and the `cause` anchor in the specialist prompts, shared with /ship through scripts/resolvers/review-army.ts; measured 141,949 / 108,523 = 1.3080. + maxSizeRatio: 1.3378, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 145,176 / 108,523 = 1.33774, rounded up (before: 1.3369). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.3369 (before: 1.3343). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.3343 (before: 1.3288). Before that: + GST-1088 lane B follow-up (raised from 1.3283 to measured, no headroom): the Done-means check/waive commands run through the host-resolved bin path instead of a bare gstack-skill-contract that isn't on PATH (fix-check finding on #3909 at 58aeee7), +56 bytes; measured 144,203 / 108,523 = 1.328778, rounded up. Before: + GST-1088 lane B (#3909, raised from 1.3261 to measured, no headroom; owner ruling via GStack Minder 2026-11-03): the gstack-skill-contract check/waive sentence that CONTRACT_CHECKER_SHIPPED=true renders (renders/GSTA-256/plan-v3.md line 74, r2-d6 "Done-means list"), +240 bytes; lane C's diet is expected to bring /review back under 1.3180; measured 144,147 / 108,523 = 1.32826, rounded up. Before: + GST-1088 lane B (raised from 1.3180 to measured, no headroom): the contract: frontmatter block and the F1 "Done means" list, contingent on Garry's gate-rev-2 answer; lane C's diet is expected to bring /review back under 1.3180; measured 143,907 / 108,523 = 1.3261. Before: + GSTA-167 variant G part 1 (plan approved by Garry 2026-10-30, autoplan-20261030161754 "Do it", which names about +0.3k prompt tokens per voice): the native adversarial and outside-voice prompts carry the blocking floor and the Severity and Blocks lines, +537 bytes; measured 1.3180 (2026-10-30). Before: 1.3136. GSTA-96 hoist merged with GSTA-92 phase 2 (2026-10-26): both owner-approved raises together, measured 142,550 / 108,523 = 1.3135. Before: Shared-code feature + critical-severity validation: 128,042 union bytes / 108,523 baseline = 1.1799; preserves content floors. + v1.91.12.0 merge of #2999 (above, plus plan-completion fallback intent and specialist checklist-by-path) with #3002; measured 1.1843 (2026-10-01). + v1.91.19.0 wave: verdict-form Codex callers with sandbox preflight (INV-1/B1), P0/P1 structured gate (B1b), bound-plan discovery and data-not-instructions guard (F1/D3), visible LEARNINGS failures (B5); measured 129,941 / 108,523 = 1.1974 (2026-10-03). + v1.91.18.0 (#3023) merged into the v1.91.19.0 wave; measured 130,025 / 108,523 = 1.1981 (2026-10-04). + Oct 6 fix wave: quota_exhausted preflight branch (#3051); measured 130,359 / 108,523 = 1.2012 (2026-10-06). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); measured 131,314 / 108,523 = 1.2100 (2026-10-06). Combined Oct 6 wave head (A and B): measured 131,966 / 108,523 = 1.2160 (2026-10-06). + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; measured 132,199 / 108,523 = 1.2182 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; measured 132,258 / 108,523 = 1.2187 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site; measured 132,233 / 108,523 = 1.2185 (2026-10-07). Merged with the guard area: measured 132,522 / 108,523 = 1.2211 (2026-10-07). + Oct 8 wave: the defensive-security framing and fixture summary mode reach the security specialist and the Red Team (stated once in the dispatch, referenced by the Red Team); measured 133,902 / 108,523 = 1.2339 (2026-10-08). + gate integrity W2 (above); measured 1.282 (2026-10-10). + Oct 10 multi-agent wave PR D2 (sideways review): sideways sweep and flake-fix number in Step 4; `Diff read` and `Undeclared behavior changes` in the review-mode Scope Check (plan-completion section); measured 140,497 / 108,523 = 1.2946 (2026-10-10). + harness lane E (2026-10-26): the lead-judgment stage in the finding merge and the `cause` anchor in the specialist prompts, shared with /ship through scripts/resolvers/review-army.ts; measured 141,949 / 108,523 = 1.3080. }, codex: { skill: 'codex', @@ -544,8 +544,8 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, // operational skill }, behavioral: 'prompt', - maxSkeletonBytes: 32_216, // GSTA-161 skill diet (2026-10-30): lowered from 52_682 to the measured skeleton, 32,216; Step 7's canary script, the Aside contract, the browser fallback and Third-Party Web Actions moved to sections/canary.md. Before: GSTA-96 hoist (2026-10-26): lowered from 74_500 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 690 bytes; measured 51,992. Before: + Aside browser contract for Step 7 canary ({{ASIDE_SETUP}}); measured 73_523 - maxSizeRatio: 1.1344, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1344 (before: 1.1318). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1318 (before: 1.1260). Before that: + #3318 Step 3.4 reads the ship policy and prints one VERSION_CHECK verdict (owner-approved raise, Garry 2026-10-24: "do it"); measured 1.1201 (2026-10-24). Before: 1.1120. + v1.81 Aside contract + gstack-browser fallback block; measured 1.077. + both untrusted-content marker formats in the browser fallback (W7f) + the readiness gate's project test command and live eval-store paths (s06 H6/H7); measured 1.1076 (2026-10-02). Re-measured 1.0943 (2026-10-03); cap = measured + 0.005 headroom. + Oct 7 wave B6: Aside rules 1, 4, 6 and 7 (offer a target-origin tab before openTab, a second sign-in wall means a tab- or URL-bound session, no early-return hint, fast [ok without the sentinel is an abort); measured 115,226 / 104,578 = 1.1018 (2026-10-07). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file; measured 117,753 / 104,578 = 1.1260 (2026-10-25). Before: 1.125. + maxSkeletonBytes: 32_273, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 32,273 (before: 32,216 on a141e94). Before that: GSTA-161 skill diet (2026-10-30): lowered from 52_682 to the measured skeleton, 32,216; Step 7's canary script, the Aside contract, the browser fallback and Third-Party Web Actions moved to sections/canary.md. Before: GSTA-96 hoist (2026-10-26): lowered from 74_500 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 690 bytes; measured 51,992. Before: + Aside browser contract for Step 7 canary ({{ASIDE_SETUP}}); measured 73_523 + maxSizeRatio: 1.1350, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 118,689 / 104,578 = 1.13493, rounded up (before: 1.1344). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1344 (before: 1.1318). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1318 (before: 1.1260). Before that: + #3318 Step 3.4 reads the ship policy and prints one VERSION_CHECK verdict (owner-approved raise, Garry 2026-10-24: "do it"); measured 1.1201 (2026-10-24). Before: 1.1120. + v1.81 Aside contract + gstack-browser fallback block; measured 1.077. + both untrusted-content marker formats in the browser fallback (W7f) + the readiness gate's project test command and live eval-store paths (s06 H6/H7); measured 1.1076 (2026-10-02). Re-measured 1.0943 (2026-10-03); cap = measured + 0.005 headroom. + Oct 7 wave B6: Aside rules 1, 4, 6 and 7 (offer a target-origin tab before openTab, a second sign-in wall means a tab- or URL-bound session, no early-return hint, fast [ok without the sentinel is an abort); measured 115,226 / 104,578 = 1.1018 (2026-10-07). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file; measured 117,753 / 104,578 = 1.1260 (2026-10-25). Before: 1.125. minUnionBytes: 91_000, // Phase 4 wave 1; estimated union ~94.9KB mustContain: ['readiness', 'merge', 'canary', 'revert', 'staging'], }, @@ -583,7 +583,7 @@ do not launch the downstream skill or open a browser.`, maxSkeletonBytes: 48_175, // GSTA-96 hoist (2026-10-26): lowered from 70_000 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 4,888 bytes; measured 43,287. Before: Phase-specific outside coverage, native fallback, and harness guard. minUnionBytes: 85_000, // measured union 86,926 mustContain: ['6 Decision Principles', 'TASTE DECISION', 'USER CHALLENGE', 'consensus', 'Restore Point'], - maxSizeRatio: 1.1799, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1799 (before: 1.1772). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1772 (before: 1.1723). Before that: GST-537 /autoplan --ceo-only (GStack Minder approved A1, 2026-10-31): sections/ceo-only.md (2,168 B) plus its Section index row, read only when the invocation names --ceo-only (test/autoplan-ceo-only.test.ts pins that a normal run never reads it); measured 119,545 / 101,979 = 1.1723 (2026-10-31), +2,276 B over main's 117,269 = 1.1499. Before: 1.1511. Four validated outside invocations replace raw CLI calls; phases keep independent coverage. + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; measured 114,297 / 101,979 = 1.1208 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; measured 114,533 / 101,979 = 1.1231 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site, and phase-close publishing each report with the `true autoplan-published ` no-op before the next phase entry in a later message (CEO-1); measured 114,320 / 101,979 = 1.1210 (2026-10-07). Merged with the guard area: measured 115,149 / 101,979 = 1.1291 (2026-10-07). + Oct 8 wave (permission cards after a background reviewer): Step 1 puts RESTORE_PATH, and so every phase artifact, in the project's git-excluded `.gstack/tmp/autoplan/`, which Claude Code reads without a card after the skill turn ends; measured 115,270 / 101,979 = 1.1303 (2026-10-08). + fix wave PR D: the prerequisite offer is skipped in spawned/headless sessions (#1958); measured 115,267 / 101,979 = 1.1303 (2026-10-07). Both merged: measured 115,126 / 101,979 = 1.1289 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + Oct 10 multi-agent wave B9/B10/B12: Phase 0 ESTIMATE and the phase-close timing record, the final gate as a gate list (`gstack-gate render/parse`, stable ids, gate_rev) and the URGENT block above the summary, offset by tightening the Phase 4 prose and the tasks-section; measured 114,980 / 101,979 = 1.1275 (2026-10-10). + Oct 10 multi-agent wave PR D1 (reality rows): the reality-rows check line in the CEO and Eng phase files, the Pre-Gate rule (unattended miss is `incomplete`, never a warning) and the two Pre-Gate cells; measured 117,042 / 101,979 = 1.1477 (2026-10-10). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file; measured 117,388 / 101,979 = 1.1511 (2026-10-25). Before: 1.149. + maxSizeRatio: 1.1804, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 120,374 / 101,979 = 1.18038, rounded up (before: 1.1799). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1799 (before: 1.1772). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1772 (before: 1.1723). Before that: GST-537 /autoplan --ceo-only (GStack Minder approved A1, 2026-10-31): sections/ceo-only.md (2,168 B) plus its Section index row, read only when the invocation names --ceo-only (test/autoplan-ceo-only.test.ts pins that a normal run never reads it); measured 119,545 / 101,979 = 1.1723 (2026-10-31), +2,276 B over main's 117,269 = 1.1499. Before: 1.1511. Four validated outside invocations replace raw CLI calls; phases keep independent coverage. + Oct 7 wave B3: the Codex probe runs as a command (select-model capture lines, check-sandbox/probe-model/run-with-timeout subcommands) instead of being sourced; measured 114,297 / 101,979 = 1.1208 (2026-10-07). + Oct 7 wave CEO-28/DX-11: review prompts ask for a severity per finding and design proposals use the proposal gate; measured 114,533 / 101,979 = 1.1231 (2026-10-07). + Oct 7 wave guard (CEO-20): conditional `run_in_background` request plus the background-run recovery at each foreground dispatch site, and phase-close publishing each report with the `true autoplan-published ` no-op before the next phase entry in a later message (CEO-1); measured 114,320 / 101,979 = 1.1210 (2026-10-07). Merged with the guard area: measured 115,149 / 101,979 = 1.1291 (2026-10-07). + Oct 8 wave (permission cards after a background reviewer): Step 1 puts RESTORE_PATH, and so every phase artifact, in the project's git-excluded `.gstack/tmp/autoplan/`, which Claude Code reads without a card after the skill turn ends; measured 115,270 / 101,979 = 1.1303 (2026-10-08). + fix wave PR D: the prerequisite offer is skipped in spawned/headless sessions (#1958); measured 115,267 / 101,979 = 1.1303 (2026-10-07). Both merged: measured 115,126 / 101,979 = 1.1289 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + Oct 10 multi-agent wave B9/B10/B12: Phase 0 ESTIMATE and the phase-close timing record, the final gate as a gate list (`gstack-gate render/parse`, stable ids, gate_rev) and the URGENT block above the summary, offset by tightening the Phase 4 prose and the tasks-section; measured 114,980 / 101,979 = 1.1275 (2026-10-10). + Oct 10 multi-agent wave PR D1 (reality rows): the reality-rows check line in the CEO and Eng phase files, the Pre-Gate rule (unattended miss is `incomplete`, never a warning) and the two Pre-Gate cells; measured 117,042 / 101,979 = 1.1477 (2026-10-10). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file; measured 117,388 / 101,979 = 1.1511 (2026-10-25). Before: 1.149. }, spec: { skill: 'spec', @@ -612,7 +612,7 @@ do not launch the downstream skill or open a browser.`, behavioral: 'prompt', maxSkeletonBytes: 36_571, // GSTA-96 hoist (2026-10-26): lowered from 58_400 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 1,606 bytes; measured 34,965. Before: + v2.0 {{ASIDE_RESEARCH}} (Aside first, WebSearch fallback); measured 57_356 minUnionBytes: 64_500, // measured union 67,430 - maxSizeRatio: 1.0853, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.0853 (before: 1.0819). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.0819 (before: 1.0746). Before that: Oct 10 multi-agent wave PR E2: the lazy gate-and-file section renders {{COORDINATOR_CONTRACT}} (the block every spawned lane inherits word for word; `gstack-autoplan contract` prints the same bytes) before the --execute spawn pipe; measured 86,614 / 81,577 = 1.0617 after the contract became a fail-closed `CONTRACT=$(…)` capture (2026-10-10). The skeleton is unchanged (57,107). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file; measured 87,656 / 81,577 = 1.0745 (2026-10-25). Before: 1.07. + maxSizeRatio: 1.0860, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 88,590 / 81,577 = 1.08597, rounded up (before: 1.0853). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.0853 (before: 1.0819). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.0819 (before: 1.0746). Before that: Oct 10 multi-agent wave PR E2: the lazy gate-and-file section renders {{COORDINATOR_CONTRACT}} (the block every spawned lane inherits word for word; `gstack-autoplan contract` prints the same bytes) before the --execute spawn pipe; measured 86,614 / 81,577 = 1.0617 after the contract became a fail-closed `CONTRACT=$(…)` capture (2026-10-10). The skeleton is unchanged (57,107). + GSTA-96 shared-preamble hoist (owner-approved raise, Garry 2026-10-25: "do it"): the union now holds sections/preamble.md, so a fresh invocation reads the pointer plus the shared file; measured 87,656 / 81,577 = 1.0745 (2026-10-25). Before: 1.07. mustContain: ['HARD GATE', 'dedupe', 'quality gate', 'acceptance criteria', 'archive'], }, 'setup-gbrain': { @@ -640,7 +640,7 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, }, behavioral: 'prompt', - maxSkeletonBytes: 38_629, // GSTA-96 hoist (2026-10-26): lowered from 60_450 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 323 bytes; measured 38,306. Before: + v1.78 AUQ spawned-trigger objectivity (explicit declaration + interactive fence); measured 60_013 + maxSkeletonBytes: 38_676, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 38,676 (before: 38,629 on a141e94). Before that: GSTA-96 hoist (2026-10-26): lowered from 60_450 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 323 bytes; measured 38,306. Before: + v1.78 AUQ spawned-trigger objectivity (explicit declaration + interactive fence); measured 60_013 minUnionBytes: 78_300, // measured union 79,139 mustContain: ['PGLite', 'Supabase', 'claude mcp add', 'read_secret_to_env', 'pooler'], maxSizeRatio: 1.07, // measured 1.051 vs the branch monolith: index + stubs + 4 STOP pointers @@ -677,7 +677,7 @@ do not launch the downstream skill or open a browser.`, }, behavioral: 'prompt', maxSkeletonBytes: 41_669, // GSTA-96 hoist (2026-10-26): lowered from 63_500 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 12,415 bytes; measured 29,254. Before: + v2.0 {{ASIDE_SETUP}}/{{BROWSE_FALLBACK}} (replaces the browse setup block); measured 61_253 - maxSizeRatio: 1.1526, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1526 (before: 1.1493). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1493 (before: 1.1420). Before that: + v1.81 Aside contract + gstack-browser fallback block (1.080 on v1.91.7.0) + the shared test value bar at 8a.5 ({{TEST_VALUE_BAR:qa}}); measured 1.094 + W1 guarded state-root resolution (`eval gstack-paths; : "${GSTACK_STATE_ROOT:?…}"`) in the Context Recovery preamble, the eureka log and each state-writing bash block; measured 1.101 (2026-09-30) + v1.91.12.0 merge of #2999 (await scope/method Reads, capture --after checkpoints, browser-only empty evidence list) with #3002; measured 1.1028 (2026-10-01). + both untrusted-content marker formats in the browser fallback (W7f); measured 1.1035 (2026-10-02). Re-measured 1.1038 after the W8 probe-loop renames (2026-10-03); the cap already matches. + the annotate-every-capture and superseded rule before materialize (a ship-exploratory-late-input run published evidence that omitted older-snapshot captures); measured 1.1064 (2026-10-03). + v1.91.18.0 (#3023) merged into the v1.91.19.0 wave; measured 93,282 / 84,049 = 1.1099 (2026-10-04). + reliability follow-ups C2 (judge-named gaps after trimming the added wording: feasible 180 s Browser Quick, PROBE_DIR's source line, stop goes to the §4 report, written JSON keeps full card values; lane S's {user}/timing fields); measured 93,311 / 84,049 = 1.11020 (2026-10-04). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); measured 94,486 / 84,049 = 1.1242 (2026-10-06). + Oct 7 wave B6: Aside rules 1, 4, 6 and 7 (offer a target-origin tab before openTab, a second sign-in wall means a tab- or URL-bound session, no early-return hint, fast [ok without the sentinel is an abort); measured 94,681 / 84,049 = 1.1265 (2026-10-07). + Oct 8 wave: shared voice directive bans "load-bearing" and adds the reply-language rule, offset by tightening the bounded closer (net +17 bytes per tier-2+ skill); measured 94,736 / 84,049 = 1.12715 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + maxSizeRatio: 1.1533, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 96,928 / 84,049 = 1.15323, rounded up (before: 1.1526). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.1526 (before: 1.1493). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.1493 (before: 1.1420). Before that: + v1.81 Aside contract + gstack-browser fallback block (1.080 on v1.91.7.0) + the shared test value bar at 8a.5 ({{TEST_VALUE_BAR:qa}}); measured 1.094 + W1 guarded state-root resolution (`eval gstack-paths; : "${GSTACK_STATE_ROOT:?…}"`) in the Context Recovery preamble, the eureka log and each state-writing bash block; measured 1.101 (2026-09-30) + v1.91.12.0 merge of #2999 (await scope/method Reads, capture --after checkpoints, browser-only empty evidence list) with #3002; measured 1.1028 (2026-10-01). + both untrusted-content marker formats in the browser fallback (W7f); measured 1.1035 (2026-10-02). Re-measured 1.1038 after the W8 probe-loop renames (2026-10-03); the cap already matches. + the annotate-every-capture and superseded rule before materialize (a ship-exploratory-late-input run published evidence that omitted older-snapshot captures); measured 1.1064 (2026-10-03). + v1.91.18.0 (#3023) merged into the v1.91.19.0 wave; measured 93,282 / 84,049 = 1.1099 (2026-10-04). + reliability follow-ups C2 (judge-named gaps after trimming the added wording: feasible 180 s Browser Quick, PROBE_DIR's source line, stop goes to the §4 report, written JSON keeps full card values; lane S's {user}/timing fields); measured 93,311 / 84,049 = 1.11020 (2026-10-04). + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); measured 94,486 / 84,049 = 1.1242 (2026-10-06). + Oct 7 wave B6: Aside rules 1, 4, 6 and 7 (offer a target-origin tab before openTab, a second sign-in wall means a tab- or URL-bound session, no early-return hint, fast [ok without the sentinel is an abort); measured 94,681 / 84,049 = 1.1265 (2026-10-07). + Oct 8 wave: shared voice directive bans "load-bearing" and adds the reply-language rule, offset by tightening the bounded closer (net +17 bytes per tier-2+ skill); measured 94,736 / 84,049 = 1.12715 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). minUnionBytes: 69_500, // measured union 70,385 // 'aside repl' pins the Aside contract; '$B goto' pins the fallback block in the always-loaded skeleton. mustContain: ['bug', 'aside repl', '$B goto', 'fix', 'Health Score Rubric', 'regression'], @@ -714,7 +714,7 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, }, behavioral: 'prompt', - maxSkeletonBytes: 35_574, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 35,574 (before: 35_298). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 35,298 (before: 34_700). Before that: + v2.0 {{ASIDE_SETUP}}/{{BROWSE_FALLBACK}} skeleton + the STOP-Read pointer; measured 29_986 + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); skeleton measured 32,532 (2026-10-06). + Oct 7 wave B6: Aside rules 1, 4, 6 and 7 (offer a target-origin tab before openTab, a second sign-in wall means a tab- or URL-bound session, no early-return hint, fast [ok without the sentinel is an abort), and the cookbook's silent-failure list and signed-in-tab recipe (the rule details live here, not in every browsing skill); skeleton measured 34,047 (2026-10-07). + Oct 8 wave: the tier-1 voice directive bans "load-bearing" and adds the reply-language rule (tier 1 has no closer to tighten: +158 bytes); skeleton measured 34,209 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + maxSkeletonBytes: 35_635, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 35,635 (before: 35,574 on a141e94). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 35,574 (before: 35_298). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 35,298 (before: 34_700). Before that: + v2.0 {{ASIDE_SETUP}}/{{BROWSE_FALLBACK}} skeleton + the STOP-Read pointer; measured 29_986 + Oct 6 fix wave A3/CEO-12: free text reaches commands only through agent-written mktemp files (mktemp block, write rule, path rebuild and refusal at each posting, prompt and brief site); skeleton measured 32,532 (2026-10-06). + Oct 7 wave B6: Aside rules 1, 4, 6 and 7 (offer a target-origin tab before openTab, a second sign-in wall means a tab- or URL-bound session, no early-return hint, fast [ok without the sentinel is an abort), and the cookbook's silent-failure list and signed-in-tab recipe (the rule details live here, not in every browsing skill); skeleton measured 34,047 (2026-10-07). + Oct 8 wave: the tier-1 voice directive bans "load-bearing" and adds the reply-language rule (tier 1 has no closer to tighten: +158 bytes); skeleton measured 34,209 (2026-10-08). + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). minUnionBytes: 36_000, // skeleton + command-list section (14,240); measured union ~38.5 KB // 'BEGIN/END UNTRUSTED EXTERNAL' pins the untrusted-content warning in the // command reference; 'aside repl' pins the Aside contract in the skeleton. @@ -733,7 +733,7 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, }, behavioral: 'prompt', - maxSkeletonBytes: 51_626, // GST-2726 (2026-11-15, Garry: bring /plan-eng-review back): the EUREKA example names /plan-eng-review again; measured 51,626 (before: 51,619 on 81fb900). Before that: GSTA-96 hoist (2026-10-26): lowered from 73_450 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 937 bytes; measured 50,685. Before: + v1.78 AUQ spawned-trigger objectivity (explicit declaration + interactive fence); measured 73_059 + maxSkeletonBytes: 51_669, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 51,669 (before: 51,626 on a141e94). Before that: GST-2726 (2026-11-15, Garry: bring /plan-eng-review back): the EUREKA example names /plan-eng-review again; measured 51,626 (before: 51,619 on 81fb900). Before that: GSTA-96 hoist (2026-10-26): lowered from 73_450 by the shared preamble that moved to sections/preamble.md, keeping the pre-hoist headroom of 937 bytes; measured 50,685. Before: + v1.78 AUQ spawned-trigger objectivity (explicit declaration + interactive fence); measured 73_059 minUnionBytes: 66_000, // measured union 73,496 mustContain: ['retrospective', '45-minute gap', 'Ship of the week', 'Praise'], }, @@ -757,10 +757,10 @@ do not launch the downstream skill or open a browser.`, // No section-loading paid case yet: a new periodic case would push the // periodic census past its 38-slice cap (test/paid-retry-supervision.test.ts). behavioral: 'none', - maxSkeletonBytes: 85_182, // /design-review --build (GST-1741 PR 2, 2026-11-07, grant: GStack Minder, 2026-11-07): --build routing, stop pointer, index row, description clause and trigger: +443 B, trimmed 130; the build-loop section is mode-loaded; measured 85,182. Before: GSTA-134 skill diet (2026-10-28): lowered from 115_947 to the measured skeleton after the outside voices and test bootstrap moved to sections, 84,869. Before: GSTA-96 hoist merged with v1.91.274.0 (#3507, 2026-10-27): the hoisted skeleton plus main's headroom of 0 bytes; measured 115,947 on the merged tree. Before: measured 137735 on the merged tree with main f27a760 (2026-10-27, #3507); 137767 (2026-10-26; 137773 on 2026-10-25 before the shorter craft-audit trigger): the Section index the external-host carve test requires, its craft-audit row, and the STOP pointer (GSTA-92; Garry 2026-10-25 'Do it' (GSTA-92)). + maxSkeletonBytes: 85_222, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 85,222 (before: 85,182 on a141e94). Before that: /design-review --build (GST-1741 PR 2, 2026-11-07, grant: GStack Minder, 2026-11-07): --build routing, stop pointer, index row, description clause and trigger: +443 B, trimmed 130; the build-loop section is mode-loaded; measured 85,182. Before: GSTA-134 skill diet (2026-10-28): lowered from 115_947 to the measured skeleton after the outside voices and test bootstrap moved to sections, 84,869. Before: GSTA-96 hoist merged with v1.91.274.0 (#3507, 2026-10-27): the hoisted skeleton plus main's headroom of 0 bytes; measured 115,947 on the merged tree. Before: measured 137735 on the merged tree with main f27a760 (2026-10-27, #3507); 137767 (2026-10-26; 137773 on 2026-10-25 before the shorter craft-audit trigger): the Section index the external-host carve test requires, its craft-audit row, and the STOP pointer (GSTA-92; Garry 2026-10-25 'Do it' (GSTA-92)). minUnionBytes: 136_100, // GSTA-134 skill diet (2026-10-28): measured union 137,602 (lowered from 143_300); before: measured union 144802 on the merged tree with main f27a760 (2026-10-27, #3507); 145037 (2026-10-26) mustContain: ['AI Slop Detection', 'Fix Loop', 'gstack-design-lint', 'Proof beside each claim', 'holds session data'], - maxSizeRatio: 1.3654, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.3654 (before: 1.3628). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.3628 (before: 1.357). Before that: /design-review --build (GST-1741 PR 2, 2026-11-07, grant: GStack Minder, 2026-11-07): --build routing, stop pointer, index row, description clause and trigger: +443 B, trimmed 130; the build-loop section is mode-loaded; measured 143,620 / 105,842 = 1.35693. Before: 1.3001, GSTA-134 skill diet (2026-10-28): measured 137,602 / 105,842 = 1.30007, rounded up (lowered from 1.3739). Before: GSTA-96 hoist merged with v1.91.274.0 (#3507, 2026-10-27): the union now holds sections/preamble.md (the hoist's owner-approved raise, decision A) on top of main's 1.3681; measured 145,410 / 105,842 = 1.3739 on the merged tree. Before: first parity guard for design-review: measured 144802 / 105,842 = 1.3681 on the merged tree with main f27a760 (2026-10-27, #3507: the design lint swap); before that 145082 / 105,842 = 1.3708 (2026-10-25; 145077 after the /ship review fixes, 145068 = 1.3707 after the line-13 reword, 145049 = 1.3705 after the post-merge /ship review, 145037 = 1.3704 after its second pass, 2026-10-26), the skeleton's growth since v1.64.1.0 plus the craft audit (Garry 2026-10-25 'Do it' (GSTA-92)). + maxSizeRatio: 1.3657, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 144,546 / 105,842 = 1.36568, rounded up (before: 1.3654). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 1.3654 (before: 1.3628). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 1.3628 (before: 1.357). Before that: /design-review --build (GST-1741 PR 2, 2026-11-07, grant: GStack Minder, 2026-11-07): --build routing, stop pointer, index row, description clause and trigger: +443 B, trimmed 130; the build-loop section is mode-loaded; measured 143,620 / 105,842 = 1.35693. Before: 1.3001, GSTA-134 skill diet (2026-10-28): measured 137,602 / 105,842 = 1.30007, rounded up (lowered from 1.3739). Before: GSTA-96 hoist merged with v1.91.274.0 (#3507, 2026-10-27): the union now holds sections/preamble.md (the hoist's owner-approved raise, decision A) on top of main's 1.3681; measured 145,410 / 105,842 = 1.3739 on the merged tree. Before: first parity guard for design-review: measured 144802 / 105,842 = 1.3681 on the merged tree with main f27a760 (2026-10-27, #3507: the design lint swap); before that 145082 / 105,842 = 1.3708 (2026-10-25; 145077 after the /ship review fixes, 145068 = 1.3707 after the line-13 reword, 145049 = 1.3705 after the post-merge /ship review, 145037 = 1.3704 after its second pass, 2026-10-26), the skeleton's growth since v1.64.1.0 plus the craft audit (Garry 2026-10-25 'Do it' (GSTA-92)). }, // ── Token-reduction Phase 4 wave 4 (v1.69.x branch): design doctrine carve ── @@ -790,7 +790,7 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, // operational skill, no plan-mode gate }, behavioral: 'prompt', - maxSkeletonBytes: 34_410, // GSTA-96 hoist merged with v1.91.274.0 (#3507, 2026-10-27): lowered from 56_200 by the shared preamble that moved to sections/preamble.md, keeping main's headroom of 366 bytes; measured 34,044 on the merged tree. Before: measured 55,834 on the merged tree with main f27a760 (2026-10-27, #3507: the Step 4 lint line); 55,786 (2026-10-25): the craft section pointer, its index row and the Step 4 craft-check line (GSTA-92); before that 55,262 (2026-09-09): the detector install offer pointer + its sections-table row (the brief itself lives in sections/detector-install-offer.md); before that 54,545 for the review-cycle trust prose, the Slop Gate's Decisions-Log clause, and the blacklist header's override sentence + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). + maxSkeletonBytes: 34_427, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 34,427 (before: 34,410 on a141e94). Before that: GSTA-96 hoist merged with v1.91.274.0 (#3507, 2026-10-27): lowered from 56_200 by the shared preamble that moved to sections/preamble.md, keeping main's headroom of 366 bytes; measured 34,044 on the merged tree. Before: measured 55,834 on the merged tree with main f27a760 (2026-10-27, #3507: the Step 4 lint line); 55,786 (2026-10-25): the craft section pointer, its index row and the Step 4 craft-check line (GSTA-92); before that 55,262 (2026-09-09): the detector install offer pointer + its sections-table row (the brief itself lives in sections/detector-install-offer.md); before that 54,545 for the review-cycle trust prose, the Slop Gate's Decisions-Log clause, and the blacklist header's override sentence + Honest Work wave merged (W2/W3 gate integrity in review/ship, W4 kernel + DONE line in every tier-2+ skill, W5/W7 in spec/ship): re-measured on the integrated tree (2026-10-10). minUnionBytes: 57_500, // Phase 4 wave 4; measured union 58,682; 78,626 on the merged tree with main f27a760 (2026-10-27, #3507) mustContain: ["Don't make me think", "Users scan, they don't read", 'The Goodwill Reservoir', 'PRETEXT API CHEATSHEET', 'Pattern 3: Text around obstacles'], }, @@ -853,7 +853,7 @@ do not launch the downstream skill or open a browser.`, }, behavioral: 'external', externalTest: 'test/skill-e2e-design.test.ts', - maxSkeletonBytes: 18_112, // GSTA-96 hoist (2026-10-27): lowered from 39_900 by the shared preamble that moved to sections/preamble.md, keeping main's headroom of 32 bytes; measured 18,080 on the v1.91.273.0 merge. Before: new skill (GSTA-92 plan item 15, Garry 2026-10-25 "Do it"); measured skeleton 39,836 (2026-10-25) + maxSkeletonBytes: 18_132, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 18,132 (before: 18,112 on a141e94). Before that: GSTA-96 hoist (2026-10-27): lowered from 39_900 by the shared preamble that moved to sections/preamble.md, keeping main's headroom of 32 bytes; measured 18,080 on the v1.91.273.0 merge. Before: new skill (GSTA-92 plan item 15, Garry 2026-10-25 "Do it"); measured skeleton 39,836 (2026-10-25) minUnionBytes: 57_500, // measured union 58,666 (2026-10-25) mustContain: ['window.__start()', 'window.__tick(dt)', 'whole milliseconds averaging 1000/60', 'Content never hides behind animation timing'], }, @@ -900,7 +900,7 @@ do not launch the downstream skill or open a browser.`, gateAfterStop: undefined, // operational skill, no plan-mode gate }, behavioral: 'prompt', - maxSkeletonBytes: 32_169, // GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 32,169 (before: 31_893). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 31,893 (before: 31_300). Before that: first carve: Section index, one STOP pointer and the vet rule on the 29.9k monolith; measured 30,781 (2026-10-26) + maxSkeletonBytes: 32_225, // GST-2774 (2026-11-14): the Claude catalog description keeps a when-to-use clause (GSTA-558: Opus 5.5 skipped /review and the planning skills when the listing said only what a skill is); measured 32,225 (before: 32,169 on a141e94). Before that: GST-2756 (2026-11-15, Garry's "Do it" to the ask-once and open-every-file habits): two Ground Rules lines (ask each question once and carry it as one waiting line; open every handed-over file and remove placeholders), +276 bytes; measured 32,169 (before: 31_893). Before that: GST-2753 (2026-11-15, Garry's "Do it" to moving the never-print-secrets rule into gstack): the shared Ground Rules preamble section (never print a whole environment, git config or process table; wait on CI with one gstack-pr-wait call), +614 bytes; measured 31,893 (before: 31_300). Before that: first carve: Section index, one STOP pointer and the vet rule on the 29.9k monolith; measured 30,781 (2026-10-26) minUnionBytes: 33_000, // skeleton + vet-the-number section; measured union 34,527 (2026-10-26) mustContain: ['performance.getEntries()', 'REGRESSION', 'Name the limiter', 'run-to-run variation', 'inconclusive'], // The v1.64.1.0 baseline measured the 37.2k monolith; the carved union is below it. diff --git a/test/shared-libs-rendering.test.ts b/test/shared-libs-rendering.test.ts index 07bde9b55..7c0a5f0c5 100644 --- a/test/shared-libs-rendering.test.ts +++ b/test/shared-libs-rendering.test.ts @@ -167,10 +167,11 @@ describe('shared-code skill distribution', () => { } }); - test('standalone catalog entry adds exactly the approved 82 bytes', () => { + // 82 approved bytes, plus 36 for the when-to-use clause every Claude catalog entry gained in GST-2774 and the YAML quotes it needs. + test('standalone catalog entry adds exactly the approved 118 bytes', () => { const skill = readFileSync(join(output, 'deslop-shared-libs/SKILL.md'), 'utf8'); const description = skill.match(/^description: (.+)$/m)?.[1]; - expect(description).toBe('Find worthwhile shared-code extractions in recent work. (gstack)'); - expect(Buffer.byteLength('deslop-shared-libs') + Buffer.byteLength(description!)).toBe(82); + expect(description).toBe('"Find worthwhile shared-code extractions in recent work. Use for: find code worth sharing. (gstack)"'); + expect(Buffer.byteLength('deslop-shared-libs') + Buffer.byteLength(description!)).toBe(118); }); }); diff --git a/test/size-ledger/ship/claude-skill-descriptions-say-when-to-use.json b/test/size-ledger/ship/claude-skill-descriptions-say-when-to-use.json new file mode 100644 index 000000000..0f19f9b91 --- /dev/null +++ b/test/size-ledger/ship/claude-skill-descriptions-say-when-to-use.json @@ -0,0 +1,4 @@ +{ + "bytes": 29, + "why": "GST-2774: the Claude catalog description keeps a when-to-use clause (Use for: ship it, create a pr, push to main)" +} diff --git a/test/skill-triggers.test.ts b/test/skill-triggers.test.ts index bc53bb03f..daa0074fe 100644 --- a/test/skill-triggers.test.ts +++ b/test/skill-triggers.test.ts @@ -41,6 +41,8 @@ describe('routerLine (router-mode targets)', () => { l => { expect(l).toStartWith('Use when asked to'); expect(l).not.toContain('Pre-landing'); }], ['no trigger sentence: the triggers list leads', 'Pre-landing PR review. Analyzes the diff.', TRIGGERS, 160, l => expect(l).toStartWith('Use for: review this pr, code review')], + ['the catalog-trimmed "Use for:" clause gives way to the full triggers list', 'Pre-landing PR review. Use for: review this pr, code review. (gstack)', TRIGGERS, 160, + l => expect(l).toBe('Use for: review this pr, code review, check my diff, pre-landing review. Pre-landing PR review.')], ['neither: the first sentence, cut at a word', 'A very long first sentence about a workflow that goes on and on and on.', [], 40, l => { expect(l).toStartWith('A very long first sentence'); expect(l).toEndWith('...'); }], ]; diff --git a/ubicloud/SKILL.md b/ubicloud/SKILL.md index e83959ac9..3999e650f 100644 --- a/ubicloud/SKILL.md +++ b/ubicloud/SKILL.md @@ -2,7 +2,7 @@ name: ubicloud preamble-tier: 2 version: 1.0.0 -description: "Run a heavy suite, build or eval on an ephemeral 16-vCPU Ubicloud VM: TTL-named VMs, a reaper, quota preflight and a hang watchdog. (gstack)" +description: "Run a heavy suite, build or eval on an ephemeral 16-vCPU Ubicloud VM: TTL-named VMs, a reaper, quota preflight and a hang watchdog. Use for: run on ubicloud, ubicloud vm, remote test run. (gstack)" triggers: - run on ubicloud - ubicloud vm diff --git a/unfreeze/SKILL.md b/unfreeze/SKILL.md index ca32654bd..5c86cd3aa 100644 --- a/unfreeze/SKILL.md +++ b/unfreeze/SKILL.md @@ -1,7 +1,7 @@ --- name: unfreeze version: 0.1.0 -description: Clear the freeze boundary set by /freeze, allowing edits to all directories again. (gstack) +description: "Clear the freeze boundary set by /freeze, allowing edits to all directories again. Use for: unfreeze edits, unlock all directories. (gstack)" triggers: - unfreeze edits - unlock all directories diff --git a/verify/SKILL.md b/verify/SKILL.md index cd9882c2b..619e582fe 100644 --- a/verify/SKILL.md +++ b/verify/SKILL.md @@ -2,7 +2,7 @@ name: verify preamble-tier: 1 version: 0.1.0 -description: "Use when coding: run the checks and tests /ship picks until green. (gstack)" +description: "Use when coding: run the checks and tests /ship picks until green. Use for: verify my change, run the checks, make the tests pass. (gstack)" triggers: - verify my change - run the checks diff --git a/write-skill/SKILL.md b/write-skill/SKILL.md index 02732b264..8615374ba 100644 --- a/write-skill/SKILL.md +++ b/write-skill/SKILL.md @@ -2,7 +2,7 @@ name: write-skill preamble-tier: 2 version: 0.1.0 -description: "Write a new agent skill or improve one without bloat: the edit ladder, levels, drift and size checks, full diets. (gstack)" +description: "Write a new agent skill or improve one without bloat: the edit ladder, levels, drift and size checks, full diets. Use for: write a skill, improve this skill. (gstack)" triggers: - write a skill - improve this skill