Repository navigation

feat(eval): calibrated matcher, fail-closed control verdicts, and a negative result - #68

Merged

ManSio merged 14 commits into

mainfrom

exp/prompt-robustness-v3

Oct 4, 2026

Commits on Oct 3, 2026

chore(docs): archive auto-synced diary copies out of the issue board
MSCodeBase Agent
committed
36801ac
fix(deps): bump urllib3 and PyJWT to clear 16 known CVEs
MSCodeBase Agent
committed
58b95b7
fix(tests): remove the cross-test race that made CI classify a healthy guard as UNPROVEN
MSCodeBase Agent
committed
ee5edd2
feat(guards): add third-party data and cross-session pre-commit gates
MSCodeBase Agent
committed
265ce2f
fix(gate-zero): run pytest under python.exe, not pythonw
MSCodeBase Agent
committed
c4fda0a
docs(registries): diary, wisdom and known-issues entries for the new gates
MSCodeBase Agent
committed
7d6366a
fix(eval): align base_repeats default with CLI and ignore .pr_work
MSCodeBase Agent
committed
22642cf
test(eval): parse CLI stdout/stderr and assert both language arms
MSCodeBase Agent
committed
46bb2ab
chore(eval): record prompt_robustness v2 live run and findings
MSCodeBase Agent
committed
31323da
feat(eval): structured matcher rules and fail-closed control verdicts
MSCodeBase Agent
committed
6f8afe9
docs(eval): record v3 matcher calibration and control-verdict rework
MSCodeBase Agent
committed
b9afd4c
docs(eval): record v3 live run - verdict column not comparable, new FN class
MSCodeBase Agent
committed
0718e5f
docs(eval): live-distribution calibration is degenerate - 16:1 noise to signal
MSCodeBase Agent
committed
08aa4e2
docs(eval): propose diary, wisdom and known-issues entries for the registry owner
MSCodeBase Agent
committed
53e9651