Repository navigation
Codex loads office-hours, plan-design-review and gstack-upgrade again, and knows when it runs unattended (GST-3161) - #4510
Merged
Conversation
…, and knows when it runs unattended (GST-3161) Setup links the three runtime-root skill references on Codex instead of copying them, so Codex lists each skill once. gstack-session-kind reports headless under codex exec. Codex subagent starts pass fork_turns none (GST-3091). Docs no longer say Codex has no hooks. Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
|
😎 This pull request was merged. |
…etup tests the new link helper The fork_turns wording grew the Codex autoplan and plan-design-review renders past their pins; shorter wording and dropping the leftover 'Agent prompt' phrasing on Codex brings both under, and the pins move down to the new sizes. Three setup tests that extract setup's functions now include _link_codex_skill_refs. Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…ill fix is macOS and Linux only Review-Finding: p1/red-team#1 reader=red-team Review-Finding: p1/specialist-batched#1 reader=testing Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
This was referenced Nov 19, 2026
Merged
capy-ai Bot
added a commit
that referenced
this pull request
Nov 19, 2026
- `gstack-artifact validate` no longer passes a run.json that gstack-autoplan never wrote. (#4494) - /ship's changed-number check no longer stops on tests that only share a word with the change (#4478). (#4496) - On Capy, /ship no longer edits or comments on pull requests as you. (#4518) - The destructive-command check stops asking about scratch deletes and catches the real damage it used to miss. (#4513) - The destructive-command check reads commands more carefully, closing gaps found after the last release. (#4525) - The Codex readiness check and the second-model review now pick the same model for a custom Codex provider (GST-3089). (#4491) - gstack's Codex hooks work when your Codex folder's path has a `(`, `&` or other shell character in it. (#4528) - Uninstalling gstack keeps your own Codex hooks trusted. (#4528) - Codex hook setup matches Codex more closely. (#4528) - On Windows, setup copies the Claude Code hook code that gstack's Codex hooks import (#4528) - /review's checklist on Codex reads shell commands more carefully. (#4528) - On Codex, gstack now suggests the right skill for a prompt and stops destructive commands before they run. (#4512) - On Codex, /review is now held once when it tries to finish without its adversarial pass or its report, as it already is on Claude Code. (#4522) - On Codex, gstack's skill suggestion now names the skill the way Codex lists it. (#4523) - On Codex, gstack's suggested prompts and docs now name skills the way Codex lists them. (#4524) - /review on Codex with gpt-6.1-sol is no longer held back for a second-model pass it already ran. (#4526) - On Codex, office-hours, plan-design-review and gstack-upgrade work again when you name them. (#4510) - gstack knows when Codex is running unattended. (#4510) - Codex reviewers start with a clean context. (#4510) - Docs no longer say Codex has no hook support. (#4510) - A command whose output was cut off no longer counts as a finished read. (#4492) - A folder marked as published data can no longer hide source code or agent files from /ship's review. (#4516) - Correction to the on-by-default helper hooks entry: the hook still routes only to /review, /qa and /verify, so skills you've installed yourself aren't suggested yet. With gstack's skills it routes 14 of 60 real bug reports (was 17), and long prompts are no longer skipped. (#4505) - /sync-gbrain's read check now works when the first page in your code source is over 64 KB. (#4497) - Free-suite CI no longer fails before any test runs when Ubuntu's package mirror is slow. (#4530) - Syncing a second repo no longer makes the first one ask for /sync-gbrain again. (#4532) - gstack's helper hooks are now on by default in Claude Code, OpenClaw and Hermes, and the prompt-routing hook fires on real bug reports. (#4503) - `gstack-pregate clear --receipt` accepts a remote gate run from before a merged release when the release only regenerated a declared release output. (#4483) - The second-model review now keeps the reviewer's own result, and reports a used-up budget as a budget stop. (#4515) - When the reviewer wrote its review to the result file it was asked for, Codex then replaced that file with its short closing message ("Done, I wrote the review"), so the review was thrown away and reported as missing its markers. (#4515) - When the provider refuses the call with HTTP 402 Payment Required (usually a metering proxy that hit its allowance, or an account out of credits or quota), gstack used to record only "exit 1". (#4515) - The pre-commit /verify check now fails a check that times out, and its checks share one time budget (GST-3201). (#4527) - /ship's review collector no longer reads a reviewer's own scope list or list of passed checks as findings (GST-3201). (#4527) - Merge-queue stack tests fail when git cannot list a PR's added test files (GST-3201). (#4527) - /ship can now review results PRs that publish raw session data. (#4511) - /ship's review pass works when the checkout you ship isn't the project's default checkout. (#4507) - A second review round no longer reports done when nothing was reviewed. (#4520) - The prompt-routing hook now suggests skills you install yourself, and no longer sends ordinary bug reports to /qa (GST-3123). (#4506) - gstack's script runner and merge-queue checks now also refuse cut-off output. (#4502) - A failed second-model review reports the cause that actually stopped it, and the runner never deletes its own prompt. (#4519) - `gstack-ship-queue ready` now fills in the trial merges and the second-model trail read instead of printing lines to fill. (#4529) - Five small ship-tool fixes. (#4529) - A ruling the owner gives while a `/ship` run is open can be relayed. (#4529) - /ship no longer kills a working test run as stalled when the test command starts its workers in their own session. (#4533) - /ship no longer stops a review because the state-machine reader started its reply the way its instructions say. (#4504) - When Claude Code runs /verify right before a commit, it now gets a quick check instead of the full fix loop (GST-3150). (#4508) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
On Codex, naming office-hours, plan-design-review or gstack-upgrade loaded nothing, because setup left a second copy of each that Codex listed twice; on macOS and Linux setup now links those copies, and re-running setup or upgrading fixes an existing install.
Skills started under
codex execnow reportSESSION_KIND: headlessinstead ofinteractive(also on slim Linux images withoutps), Codex reviewers start without the parent's conversation (fork_turns: "none"), and the docs stop saying Codex has no hooks.Nothing to do after merge; Codex users pick it up on their next upgrade.
What changed
SKILL.mdunder its skills folder and skips a symlinked one.create_codex_runtime_rootcopiedgstack-upgrade,office-hoursandplan-design-reviewinto~/.codex/skills/gstack/so other skills can read them by path; those copies were second listings of skills Codex already lists fromgstack-<name>/, and naming a skill that is listed twice injects nothing. The new_link_codex_skill_refswrites them as links, so the paths still resolve and each skill is listed once. The runtime root is rebuilt on every setup run, so an existing install is fixed by./setup --host codexor/gstack-upgrade. Windows (copies, not links) still lists them twice, and the changelog says so.bin/gstack-session-kind). Codex gives tool shells the same environment in the TUI and in exec (CODEX_THREAD_ID,CODEX_CI=1, no TTY), so the helper reads the subcommand of the nearestcodexprocess:exec,eorreviewmeans headless; the TUI runs tool shells undercodex app-serverand stays interactive. It reads parent pids from /proc on Linux (falling back tops), only walks the process tree whenCODEX_THREAD_IDis set, and Conductor's markers still win.hosts/codex.ts). The inlinespawn_agentcall used by the design reviewers, /autoplan's phases and /context-restore now passesfork_turns: "none"(Codex defaults it toall). /ship's audits already did.docs/skill-routing-and-contracts.mdsays Codex runs hooks but gstack hasn't ported its three there;docs/unattended.mddescribes the Codex session-kind rule.Why
"
$office-hoursstarted without its instructions" on Codex, and acodex execrun asked questions in prose that nobody would answer. Codex users now get each skill listed once and loaded when named, and unattended Codex runs stop with a clear BLOCKED line instead.Blast radius
Every gstack workflow reads
gstack-session-kind, so a wrong answer would change question handling everywhere; the new Codex branch runs only whenCODEX_THREAD_IDis set and a codex ancestor runsexec,eorreview, and any read failure falls back to the oldinteractiveanswer (walked failure path, tests ran). The setup change touches only the three Codex reference copies, which other skills read by path; the links resolve to the committed.agents/skillsrender (live Codex 0.170.0 install and upgrade checked).Pre-Landing Review
Two full passes on the gate-class review plan (specialists for testing, security and state machines, red team, a native adversarial reader, a second-model review on GPT-6.1 Sol and a targeted auth reader). Pass 1 raised two informational findings, both fixed in d7f2316: a testing gap (Windows setup still copies the three files, so the changelog now says macOS and Linux) and one red-team finding (security finding, details recorded privately). Pass 2 on the fix range was clean; the second-model review returned CLEAN and the targeted reader found nothing. No blocking findings.
Review notes
Below the blocking floor (advisory, informational, test-gap or wording items); recorded here and not fixed in this PR.
bin/gstack-session-kind:105If an ancestor exits between the -r check and the read, bash prints a No such file line to stderr; result stays interactive.bin/gstack-session-kind:105The PPid assignment can return 1; safe today only because the function runs inside an if condition under set -e.bin/gstack-session-kind:78Comment says no signal on Git Bash, but MSYS2 exposes /proc too; behavior still safe (native codex.exe shows as PPid 1).Exploratory QA
Functional smoke on
bin/gstack-session-kindwith fake codex parents, rerun after the fix: codex exec → headless, app-server → interactive, a--no-daemonprompt starting with "exec" → interactive, and codex exec with nopson PATH (direct and through an intermediate shell) → headless. 5 of 5 pass. Limit: fake parents, not a live Codex CLI (the live checks are under Tests).Eval Results
The eval selector fell back to full coverage because
setupis a shared runtime input; it selected no quality judges and left its 145 behaviors to scheduled and release coverage, so no paid eval ran for this PR.Plan Completion
Plan completion audit: not run (no plan is bound to this branch and no docs/designs/ file matches). Fix: add "Plan: " to the PR body, or run /autoplan.
TODOS
No TODO items completed in this PR.
Documentation
Status: current. The docs already match the change, so nothing was edited.
Scope: 15 committed files from base 5afef04 to d7f2316, with no staged, unstaged or new files. The change makes Codex report
headlessundercodex exec,codex eandcodex review, links Codex's nested office-hours, plan-design-review and gstack-upgrade SKILL.md files instead of copying them, starts Codex reviewers withfork_turns: "none", and corrects the Codex hooks row in the docs.Documentation health
bin/gstack-session-kind(Conductor wins, then the Codex check, then CI)setup, which registers no Codex hooksCoverage debt: the
gstack-session-kindCodex detection has a reference entry and an explanation in docs/unattended.md, but no how-to. The non-Linux limit is not in user docs. Outside Linux,psjoins arguments with spaces, so a TUI started with--no-daemonand a prompt whose first word is "exec" would read as headless. Only the script's comment covers this._link_codex_skill_refsis internal to setup and needs no docs.Diagram drift: none. No diagram names the changed parts.
Tests
bun test test/gstack-session-kind.test.ts: 20 pass, 0 fail; the new no-pscase fails on 16b060d and passes on d7f2316.Codex CLI 0.170.0, isolated
CODEX_HOME,./setup --host codex:codex debug prompt-input)$office-hoursinjects the skill<skill><name>office-hours</name>…, about 12k tokens)gstack-session-kindrun from a real Codex tool shell:codex exec→ headless (direct and nested), Codex TUI → interactive, either withCONDUCTOR_WORKSPACE_PATH→ interactive.New tests: Codex exec, alias, global flags and
review→ headless; TUI app-server, resume and a prompt starting with "exec" → interactive; Conductor and missingCODEX_THREAD_IDnegative controls. A conformance test models Codex's listing rule and fails on main's layout, plus an upgrade test from the old real-copy layout. A render test fails on main's tenspawn_agentsites withoutfork_turns.Agent notes
Agent notes
Lane GSTA-637 for GST-3161 and GST-3091, from the GSTA-635 audit and GEVA-555. Out of scope: porting hooks to Codex, reasoning effort defaults, lib/skill-route.ts. Headless keeps its BLOCK-on-question rule, so a fresh-state
codex execnow stops with BLOCKED at the first-run telemetry question instead of asking it in prose; a parent that seeds config or setsGSTACK_SESSION_KIND=unattendedskips it.CI notes: the pull_request free-tests run 44471290795 was green but could not carry (tree_mismatch, base moved), so free-tests.yml was dispatched on the branch as run 44473881777. Its attempt 1 failed only free-suite (6), where "Install Playwright Chromium" timed out after 5 minutes (infrastructure, not a test); that one job was rerun once and attempt 2 passed. Earlier, native-arm-setup at 16b060d failed once on an Ubuntu mirror sync and passed on rerun.
Refs GST-3161, GST-3091
Approvals
No approvals recorded for branch codex-unattended-runs-and-duplicate-skills.
Plan: exempt: bug fix (--kind fix)
Plan check: exempt bug fix
Gate edits: none detected (GATE_SCAN tagged=0 at d7f2316)