ci: run the new Trust v2 / judge test suites in node-tests - #34
Merged
Conversation
test.yml used a hardcoded file list and never executed trust-scoring, adversarial-holdout or adversarial-judge — wire them into the TS suite so the Trust Score v2 + judge/calibration logic is actually CI-covered.
A judge error on a golden case was counted as a 'fail' verdict, inflating disagreement and polluting Cohen's kappa. Skip errored cases, count them as 'skipped' in the calibration record, and 502 only if every case errored.
…y-Skills
- docs/product/CYBERSEC-SKILLS-IMPORT.md: licensing, dedup, skill/playbook
routing, 3 proposed packs (SOC, Cloud/K8s, Red Team), and the upskilling path
- content/skills/{sigma-detection-engineer,pcap-triage-analyst,
kubernetes-security-auditor}.yaml: converted to our schema, Apache-2.0
attributed, ready for validate -> sync -> SkillForge
- NOTICE: Apache-2.0 attribution for the upstream corpus
… schema bar)
The schema requires >=2 examples; 38 shipped skills had a single placeholder
('<fill with a realistic task>'), so the content gate failed the moment any
content PR triggered it. Replaced each with two realistic, domain-coherent
examples (gtm-*, gx-*, doc-skill-*, cloud/devtools, ML/RAG, ads, video).
Also marks pcap-triage-analyst x_security_research (legit attacker-pattern prose).
validate:content + audit:skills now pass.
criptogus
marked this pull request as ready for review
May 29, 2026 23:09
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
While verifying the app, I found CI (
.github/workflows/test.yml) runs a hardcoded list of test files rather thannpm test. The three suites added across #31–#33 —trust-scoring,adversarial-holdout,adversarial-judge— were wired intopackage.json'stest:tsscript but never executed by CI. So the "green" checks on those PRs did not actually run the new tests.Fix
Add the three suites to the TypeScript
node:teststep so the Trust Score v2 math, the deterministic holdout split, and the LLM-judge ensemble + Cohen's κ logic are CI-covered on every push/PR.Verified locally
trust-scoring+adversarial-holdout+adversarial-judge: 26/26 passtrust,release-signing): 5/5 pass.tsfiles passnode --experimental-strip-types --check(syntax)Note: full
bun install/ typecheck / build can't run in this sandbox (npm mirror is network-restricted, 403), so the broader build is validated by CI/Lovable where deps install.https://claude.ai/code/session_01BnYrQNoKNNxPCSmV2suWna
Generated by Claude Code