mirror of
https://github.com/garrytan/gstack.git
synced 2026-11-08 09:48:20 +01:00
Garry Tanandcapy-ai[bot]
56a800da98
* feat(runaway-stopper): detect, stop and prevent runaway threads and subagents (GST-1279) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix(runaway-stopper): autoplan review fixes: input validation, exact /ship task-row meter at 100M, rotation hand-off of subagents and PR watches, bounded fetch program Skill-Size: 0→6747 bytes (SKILL.md.tmpl) Evidence: plan docs/evals/runaway-stopper-prereg.md measure_by 2026-11-24 Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * docs(ship-budget): describe the task-row reading without a literal thread_id this Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix(runaway-stopper): ship review pass 1: DEFER holds rotation and a self-stop only, working-only defer, subagents file validation, scan exit kept, DRIVER and factory.md within their size budgets Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix(runaway-stopper): ship voices: list limit clamp, allSettled fetch, waiting /ship and shipping descendants defer, unknown flags refused, GSTACK_SHIP_TASK_TITLE implies the task-row meter Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix(ship): leave DRIVER.md unchanged; main left 21 bytes under its 12 KB cap Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> --------- Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
3.4 KiB
3.4 KiB
How we work: gstack's factory on Capy
What seven Captain crews agreed on. Your crew's CREW.md and recorded approvals decide who may merge, update a base, spend or touch production; urgency changes priority, never permissions or gates. With no Captain, ask the person; never merge without their word. Story: capy/README.md; examples: org Drive software-factory/README.md.
- Plan, then a yes. New functionality starts with /autoplan sized to the work; the build waits for "do it". [SKILL.md: Plan, approve, ship]
- Someone else merges. Run /ship to a receipt on your final pushed head, send one READY quoting gstack's verify output verbatim, and stop. The merge owner merges on green and assigns versions. [docs/captain.md: The READY contract]
- Block only on the floor. A finding blocks if it would fail a test, break production, or lose data or secrets; the rest goes in the PR body or next wave. [docs/review-policy.md]
- Fix waves. When main goes red, hold /ship while one owner finds the root cause; bring main in once, as your rules allow. Fold deferred findings and test follow-ups into one PR. [docs/triage.md]
- Severe first, same gates. A severe user-facing bug jumps the queue with a short plan; receipt, checks and merge owner stay.
- Trade findings, not small talk. Another crew's finding goes to its Captain, through yours, with evidence.
- Measure before you act. Run the real system on the merged tree you ship; a real run outranks a reviewed plan. Each number names its run, commit or uncapped query; say what you did not verify.
- Evals decide, and losses get reported. Preregister bars, cap and stop rule before paying [/eval-plan]. Cheap proof (a keyless fake, a smoke) first. Controls: no-skill and placebo arms, clean decoys. Report use and pass apart; quote spend from the meter. A new part needs proof it helps or a dated test plan [ETHOS.md: 9. Earn Its Place].
- Flakes: one rerun, a ledger, a daily fix. Only the policy's rerun, never for a harness or infra red; a second red is real; never loosen a test. [docs/captain.md: The daily flaky pass]
- Quiet. Message on a decision, a blocker, a READY or a finished phase. A broadcast needing no reply goes in the Captain's notes; never reply just to acknowledge.
- Ask once, plainly. One ask per decision, explained like the owner is five: your Captain's name, then a lettered line each: "A:
hit problem p, which happens when h. We could do 1) x, 2) y or 3) z. Recommendation x because why. A1 to take recommendation." "do it" takes every recommendation; "A2" takes option 2. Owner tasks: the exact link (in Capy: path and field), numbered steps, the reply word. Never re-ask a standing rule; check what you can yourself. [docs/captain.md: The ask format] - Fresh threads, closed out. One thread per unit of work; past 500M tokens, rotate [/runaway-stopper]. Crew threads never create threads: ask your Captain (Captain to Captain across projects), or report one you made by code for crew.add. Its Captain archives only its crew's threads; threads outside every crew are the owner's, even its automation's. Size reviews to the diff; read the meter. [docs/ship-budget.md; docs/captain.md: New threads join the crew]
- The owner's model preference.
gstack-models resolve --role agentpicks agent models, except cross-family voices and eval subjects. [docs/model-policy.md: Your model preference]