Repository navigation
AGENTS/spec: astra was not short; decide 'short' by dry-run, never from config - #29
Conversation
…ch dry-run, never from config
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: 64e51e25-b0e6-43d5-81f8-4963c575db40) |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
… per-window fallback, no availability claim)
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: e305b1b5-62a3-4bf8-8dca-961ddb200506) |
Correction (lead mistake, 2026-10-31): the lead inferred "astra short at 45%" from the 60%
codex:frontierdefault in djbclark-ade's config and routed every review to Fable or Argon. The live estimate comes from measured history (about 2% of the weekly window per astra review); a dry-run on 2026-11-01 admitted astra at 41%. AGENTS now requires deciding 'short' withacp-dispatch ... --dry-run(the instruction fix, home AGENTS item 11); the spec header no longer states the false premise. Same change in frdminc/nix2cf#3.Review: astra (gpt-6-astra high,
acp-dispatch codex --kind frontier, read-only; admitted at 41% Codex weekly). Verdict: approve with changes. Applied: F1 dry-run readswould refuse:, exits 0 and skips preflight, so the real dispatch must still pass; F2 the estimate is per window from comparable measured history (14 records, scalar and weekly 2 points; the 60% fallback applies per window under three measurements); F3 the claim that astra was available on 10-31 is removed: only that the 45%-vs-60% comparison did not establish shortness.Note
Low Risk
Documentation-only changes to agent policy and a spec preamble; no runtime, auth, or deployment behavior changes.
Overview
Corrects standing-rule guidance after a 2026-10-31 lead mistake: astra was not established as short by comparing Codex’s 45% remaining to the configured 60%
codex:frontierfallback.AGENTS.mdreplaces that inference with an explicit procedure—run the intendedacp-dispatchcommand with--dry-run, readwould refuse: class=...(not exit code alone), and note that dry-run skips preflight so a clean dry-run still requires a real dispatch. It also documents how the frontier estimate uses measured history (example: 2026-11-01 astra high effort, 14 records, ~2% scalar/weekly) and when the 60% fallback applies per quota window (fewer than three measurements for that window).docs/architecture/schema-v4-reconciliation-2026-10-31.mdupdates the spec header so Argon’s cross-model seat is no longer justified by “astra short at 45%”; it states the same correction and that an astra review was admitted at 41% on 2026-11-01.Reviewed by Cursor Bugbot for commit 32e5b3f. Bugbot is set up for automated code reviews on this repo. Configure here.