Repository navigation
Findings: lint outcomes from packages[].rules, inspect-evals-lint 0.9.0, IESC001 - #14
Merged
Merged
Conversation
…t-evals-lint 0.9.0; classify IESC001 Lint writes `outcomes` only for rules that passed or skipped, so a rule with diagnostics never had an Outcome: runs_df always showed outcomes_fail = 0 for lint, rule coverage never named the failing rule, and a run whose other rules all skipped read as a skipped run and made the sweep exit 1. Outcomes now come from `rules` (one per rule that ran, at its worst status, with the diagnostic count) when present, falling back to `outcomes` for documents written before 0.9. The pin moves to 0.9.0. Its new IESC001 sandbox_privileges rule maps to environment/major rather than the harness/minor default. Fixture: real 0.9.0 output for scicode, root path redacted. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
MattFisher
changed the base branch from
dev/integrated-audits
to
feat/findings
September 30, 2026 04:24
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Independent of the milestone-1 slices; small change on dev.
Bug. inspect-evals-lint writes
outcomesonly for rules that passed or skipped. A rule with diagnostics appears only underpackages[].rules. The adapter readoutcomes, so a failing rule never had anOutcome:runs_dfalways reportedoutcomes_fail = 0for lint, rule-level coverage never named the failing rule, and a package whose remaining rules all skipped read as a skipped run and made the sweep exit 1. Verified on 0.9.0 against SciCode: rules 19 pass / 8 skip / 1 warn, outcomes 19 pass / 8 skip.Fix. Outcomes come from
ruleswhen present, one per rule that ran at its worst status (warnandfailmap tofail, as before) with a "N diagnostics" message; documents withoutrulesfall back tooutcomesunchanged. Findings are unchanged.Also. The pin moves from 0.7.0 to 0.9.0. Its new
IESC001(sandbox_privileges: privileged mode, host mounts, extra capabilities in a compose file) is classified environment/major instead of the harness/minor default.Tests: real 0.9.0 output for SciCode as a fixture; a failing rule gets an outcome and
outcomes_failcounts it; a run with only failing and skipped rules is not a skip; the 0.7.0 fixture still parses viaoutcomes. Trial on SciCode with the new pin:IEBP007 | fail | 2 diagnostics,outcomes_fail = 1, run not skipped. Gate: pre-commit clean, basedpyright 0, 108 findings tests.🤖 Generated with Claude Code