Merge main and repair review eval regressions

Preserve the upstream frontier-model defaults while retaining configured Claude
Code models for outside reviews. Repair fixed-scope mode selection and empathy
coverage counting, restore merged approval gates, and keep static scans out of
workspace archives.
This commit is contained in:
Garry Tan committed 2026-09-09 16:32:34 +00:00
commit 7b08fbfe56
147 files changed
+3244 -1707

No files matched your search

+13
View File
@@ -1056,6 +1056,19 @@ This is my **second opinion mode**.
`/codex` brings OpenAI Codex CLI to review the same diff independently. It is available on every harness except Codex itself. External harnesses install it as `/gstack-codex`. Compare its findings with the native review to distinguish corroborated findings from issues only one reviewer caught.
gstack-owned Codex calls default to `gpt-6-astra`, including resumed consult
sessions. Set `GSTACK_CODEX_MODEL=<model>` to change the default, or name a
model in your request to override it for that invocation. Generated commands
pass the selection through `-c model=...`, overriding the CLI's configured model.
Native review also sets `-c review_model=...` to that selection, overriding any
separate review-model pin.
On Codex hosts, the Claude outside-voice skill is `gstack-claude-code`. Its
review, challenge, and consult calls preserve Claude's configured model.
`GSTACK_CLAUDE_MODEL=<model>` supplies an explicit override, including resumed
sessions; a model named in your request takes precedence. Harness routing is
independent of model selection.
### Three modes
**Review** — run `codex review` against the current diff. Codex reads every changed file, classifies findings by severity (P1 critical, P2 high, P3 medium), and returns a PASS/FAIL verdict. Any P1 finding = FAIL. The review is fully independent — Codex doesn't see Claude's review.