Repository navigation
Add OTel GenAI compatibility matrix - #19
Merged
Merged
Conversation
imran-siddique
added a commit
that referenced
this pull request
Sep 25, 2026
…lusterFuzzLite (#65) `finalize_trace` took `runtime.measurement` from `snapshot.chain_digest` and derived the appraisal from `snapshot.entries`, and never checked that one still hashes to the other. `EvidenceSnapshot` is a public dataclass holding mutable dicts. Seal a run holding a deny, set `entries[0].event["decision"] = "allow"`, and the old code signed an affirming record under a measurement that commits to the deny. Fixes: - `trace_adapter.py` and `trace-finalizer.ts` recompute the chain (sequence, previous digest, entry digest, event_id, run_id, chain_digest) and refuse on mismatch. A malformed event in a consistent chain is now `TraceFinalizationError`, not `KeyError`. TS shares the digest through a new internal `evidence-digest.ts`, not re-exported. Four Python tests in `test_trace_adapter_refusals.py`, one TS test. - `validation.py`: 5,000-deep nesting raised `RecursionError`, a non-string key raised `AttributeError`, and NaN or Infinity passed. All three are now `EventValidationError`, with depth capped at 32. `test_evidence.py` expected NaN to fail at the chain; it now fails one step earlier, which matches ajv on the TS side. - `adapters/agt.py`, `agt_audit.py`, `agt_approval.py`, `agt_data.py`: a list or object where a decision, kind, outcome or classification belongs raised `TypeError: unhashable type`; `latency_ms=1e303` raised `OverflowError`. Both are `ValueError` now. The fuzzer found the first on its first run. Scorecard FuzzingID (#19): `.clusterfuzzlite/` with three targets (event validation and every projection plus inbound propagation headers; OPA and AGT policy adapters; sealed evidence to signed TRACE, including a post-seal edit that must be refused). Build installs from `requirements/test.txt` with `--require-hashes`. Actions are pinned to the SHAs agent-manifest uses, address sanitizer only, 45 minute jobs. SAST #22: `codeql.yml` already runs on push to main and every pull request with no path filter, so this clears as new commits are scanned. Verified: 128 Python tests pass, 2 skipped (was 114); TS `npm run check` 42 pass; schema, version, OTel and TS schema gates pass; wheel and sdist smoke pass; actionlint clean. Atheris will not build on Windows, so each target ran 4,000 inputs through a stand-in provider, and each was bundled with PyInstaller using the `build.sh` flags and run. That bundle run found the missing `--collect-data=agentrust_trace`. With the chain check disabled, the TRACE target caught the edited entry on 38 of 38 signed runs. Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <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.
What changed
a685613a207a580163353b8e48a7ad88967e7b42Why
The upstream GenAI conventions are Development status, have moved to a dedicated repository, and do not yet publish a schema URL. Adopters need a precise, reviewable statement of compatibility that cannot silently drift as either project changes.
Impact
Integrators can keep native OTel GenAI spans and metrics while adding AgentTrust governance facts without duplicate or falsely equivalent signals. CI now rejects unclassified projection changes.
Validation
python tools/check_otel_compatibility.py(16 attributes, 8 metrics, 6 event families)python -m unittest discover -s tests -v(87 tests, 20 subtests)python -m pytest -q(87 tests, 20 subtests)python -m compileall -q srcpython -m build