This registry records reported Concordia conformance runner runs and single-record recomputes. A row is a measurement with a linkable method and date. It does not approve, rank, or endorse an implementation.
| Implementation | Language | Author | Scope | Result | Method | Date | Link |
|---|---|---|---|---|---|---|---|
| Concordia reference runner, first-party | Python | Erik Newton | runner | [SUMMARY] positive=52 mutation=1484 canary=5 ok=1541 fail=0 |
Clean-room runner over conformance/vectors/manifest.json using RFC 8785, PyNaCl, JSON Schema, and the Python standard library. Reported result reads no conformance/diag/. |
2026-08-05 | conformance/reference-runner/runner.py |
| Concordia reference runner JS, first-party | Node.js | Erik Newton | runner | [SUMMARY] positive=52 mutation=1484 canary=5 ok=1541 fail=0 |
Runner over conformance/vectors/manifest.json using Node.js, node:crypto, and Ajv draft 2020-12. Reported result reads no conformance/diag/. |
2026-08-05 | conformance/reference-runner-js/runner.mjs |
AgentID (haroldmalikfrimpong-ops) |
Not reported | AgentID (haroldmalikfrimpong-ops) |
single-record recompute | FulfillmentAttestation digest matched sha256:47ec4298e210d3aa18832b30f8cc087b84bfebf1f664eced187918de085bf508; Ed25519 signature valid; tamper rejected; privacy shape held; charge_ref byte-matched. |
Third-party RFC 8785 JCS over the attestation minus signature, then SHA-256 and Ed25519 verify against public_key_b64url. |
2026-07-20 | A2A #1920 comment 17706287 |
The two first-party runner rows execute the same manifest to the same summary line. Divergence count: 0 vectors.
decision-object-v1 is the smallest profile: RFC 8785 JCS plus SHA-256. It is roughly 20 lines in most languages. Add profiles one at a time against the vectors.
Run the full suite or a clearly named subset with your own runner. Publish the summary line and method somewhere linkable, including whether the run read conformance/diag/ before producing its result. Then open an implementation report with the GitHub issue template.
First-party rows are labeled as first-party. Other rows should name the author, covered profiles, method, date, and result. A registry row records what was measured at that link on that date.