chore(tooling): re-vendor califa v0.9.2 — refuse logged-out eval runs - #36
Merged
Conversation
A logged-out `claude -p` exits 0 with a synthetic assistant event
("model": "<synthetic>", "error": "authentication_failed"); califa's
trigger harness scored it as a completed non-trigger, so an unauthenticated
eval run published precision=1.000 / recall=0.0 instead of refusing. v0.9.2
maps that signature to an infrastructure failure, tripping the existing
EvalIntegrityError collapse guard.
Diff vs v0.9.1: skillcard/harness/trigger.py + pyproject version only.
Byte-identical to the v0.9.2 tag; no card content_hash or scorecard SVG moves.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
CI's editable install regenerates it; the committed 0.9.1 copy tripped the scorecard drift guard. Co-Authored-By: Claude Fable 5 <noreply@anthropic.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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Re-vendors califa-cards v0.9.2 (tag
v0.9.2, commit867541b), a measurement-integrity fix to the trigger eval harness.Why
A logged-out
claude -p --output-format stream-jsonexits 0 and emits a synthetic assistant event ("model": "<synthetic>", top-level"error": "authentication_failed"). The stream decider scored that as a completed non-trigger, so a fully unauthenticated eval run passed the collapse guard and published precision=1.000 / recall=0.0 instead of refusing (confirmed live 2026-07-02). v0.9.2 maps that signature to an infrastructure failure via a newAUTH_FAILUREsentinel, so the existingEvalIntegrityErrorguard refuses the run and writes nothing.Scope
tooling/califa/skillcard/harness/trigger.py+ vendored pyproject version — the only code delta vs v0.9.1tooling/califa/SOURCEpin note updatedgit archive v0.9.2)content_hashmoves, no scorecard SVG changes —make checkgreen locally (lint, cards, scans, index, readmes)🤖 Generated with Claude Code