Repository navigation
Conversation
Add the initial Agent-side TCK integration infrastructure: probe agents for protocol v1 and experimental v2, a runner that pins TCK revisions and verifies report provenance, a CI job that uploads reports even when the v2 gate is not yet green, and docs describing local usage. The v1 lane is a CI gate (conformant, 31/31 mandatory PASS). The v2 alpha.5 lane reports known failures in version negotiation and batch responses; it is not promoted to a gate until those SDK behaviors are fixed.
PsiACE
force-pushed
the
acp-tck-evaluation
branch
from
November 4, 2026 19:31
a85be54 to
05f831c
Compare
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Adds the Agent-side ACP TCK test infrastructure so protocol conformance is verified against pinned external TCK revisions, not just the SDK's own unit tests.
scripts/run_tck.py— pins TCK revisions, runs both suites, and validates report provenance (protocol version, schema revision, agent command) before judging results.tests/tck/agent_v1.py/agent_v2.py— deterministic stdio probe agents using this checkout's public SDK entry points (acp.run_agent,acp.experimental.v2.run_agent); they supply application state only, never hand-rolled JSON-RPC framing..github/workflows/tck.yml— CI job checking out both pinned TCK commits and uploadingbuild/tck/*artifacts even when a gate fails.docs/tck.md(plus README/mkdocs links) — local run instructions and lane status.v1 gate
TCK commit
b7f40fb3(schemabeac4ba9): verdict conformant, 31/31 MANDATORY PASS. Single ADVISORY failureACP-SESSION-004is the MCP-connection check that the probe intentionally leaves to application code, not an SDK transport issue.v2 alpha.5 observation
TCK commit
fa00ccdc(schema9af0e9f7, 2.0.0-alpha.5): verdict not conformant, with known failures:ACP-INIT-003,ACP-INIT-201,ACP-INIT-202(version negotiation),ACP-JSONRPC-001(batch responses)ACP-BATCH-203/204/205,ACP-JSONRPC-005(batch per-entry handling)v2 runs in CI and reports but is not promoted to a gate until version-negotiation and batch behaviors are fixed in the SDK.
Validation
make check— clean (ruff, prek, ty, deptry, lockfile)make test— 491 passed, 1 skippedscripts/run_tck.py --tck-v1-dir <b7f40fb3> --tck-v2-dir <fa00ccdc>— exit 0; v1 conformant, v2 known failures above