Conversation
This was referenced Sep 13, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
Default local-browser tasks advertised a CDP URL but did not register browser MCP tools; Kernel tasks already did. Register the same pinned Playwright MCP server against the existing recorded browser for both modes, so Harbor agents that consume task MCP configuration (including stock Claude Code and Codex) receive browser tools by default.
The full V2 loader test now covers local and Kernel modes and checks that Harbor preserves the MCP command and CDP target. Record installed/documented versions without skipping a successful version report on newer Harbor releases.
Includes a source-linked compatibility assessment in
docs/harbor-compatibility-2026-09-13.md: runtime support versus upstream acceptance, current PR ownership, provider cleanup defects, and the oracle/parity decision still required for live websites. The documentation retains the Harbor 0.22.0 run baseline and identifies 0.23.0 coverage specifically as loader validation. Includes the required changelog entry.Corpus
Test plan
uv run --frozen pytest -q— 283 passed, 3 skipped.uv run --with harbor==0.22.0 pytest tests/test_harbor_version_compatibility.py -q— 3 passed; full V2 corpus, local + Kernel.uv run --with harbor==0.23.0 pytest tests/test_harbor_version_compatibility.py -q— 3 passed; same coverage.uv run --frozen pyright— no errors.git diff --check.Docker is not installed on this host, so no browser/provider end-to-end or model/parity experiment is claimed. Agents that do not consume task MCP configuration still need their own CDP integration. This does not claim browser-policy enforcement or official Harbor dataset registration.
Related issues
Fixes #357. Related to #331; follow-up findings tracked in #358 and #359.
Independent of #353 (schema/staging/judge status) and #346 (readiness); neither PR is folded into this change. Respects the decisions to defer mandatory CI (#350) and image publication (#331).
@Perry2004 please review the local-browser MCP default, the loader coverage, and the compatibility assessment's remaining acceptance criteria.