[antonwolfy]

Summary

Three test gaps from the reviews of #71, #153 and #105: in each, a mutant of the new code passed the full suite. The authors fixed the rest of each thread; these parts were left. Only tests change.

Changes

• feat(repo): repo budget, what a repository puts into an agent's context #71 (thread): the second escape check in GitTree.resolve (source.py:229). The new test uses docs -> real, then docs/../../x.md, and expects path escapes the repository. With the check disabled, the read failed later as links to nothing: ../x.md, and the suite passed.
• feat(repo): record whether the agent used what a patch: arm added (#41 D4) #153 (thread): repo run records uptake for a patch that adds only an MCP server, and for one that adds only a file an agent can use. The patches are each half of the existing KIT_PATCH. Gating uptake on the files alone (if uptake_of[0]) or on the servers alone passed the suite.
• refactor(novelty): one flag-to-setting resolver (RunSettings); thin the novelty CLI #105 (thread): catalog_competitors keeps each card's description, which is what the model chooses by. Building the cards with description="" passed the suite.

Risk

None: tests only. No changelog fragment, since nothing under dokimi/dokimi/ changes.

Testing

On a Linux review host, at 8fc3cce (on main 506f7ba):

• make ci-local: 29 passed, 0 failed, 7 skipped (the Docker catalog smoke, test-latest, the cli-surface packaging step, conda-forge-deps without micromamba, docs deploy, the pip-audit report, semgrep). 7,384 tests passed, 1 skipped.
• Each new test kills its mutant:
  • if False: for the escape check: 1 failed in test_repo_budget.py;
  • if uptake_of[0]: and if uptake_of[1]:: 1 failed each in test_repo_uptake.py;
  • description="": 1 failed in test_novelty_routing.py.
• Rebased onto main 92d92d3 (test(llm): pin provider-call budgets for judge, panel, answer and validation paths #170, tests only) before opening; with the ten follow-up branches merged together, the full suite (7,447 passed, 1 skipped), the growth check, mypy and ruff pass.

Checklist

• Tests added/updated and pytest passes
• ruff, mypy, and schema checks pass
• Docs updated (or N/A) — see the docs-drift checklist in CONTRIBUTING.md
• No ad-hoc run artifacts committed
• New files carry the SPDX license header (or are exempt in .github/.licenserc.yaml)

[vchamarthi]

@exai-bot /review

[exai-bot]

Note

AI-generated review of c0b5b46c63c4, requested by @vchamarthi, using the default review skill. It can be wrong: verify a finding before acting on it. React with a thumbs up or down on a finding to tell the bot whether it helped.
Model us.anthropic.claude-sonnet-5-5, effort default, 9 of 80 turns, 0.6 of 40 minutes, 269k in / 3k out tokens.

Tests-only change that pins three previously unpinned checks: the second escape check in GitTree.resolve, repo run uptake for a patch with only a server or only a file, and the description kept on catalog_competitors cards. I found no defects. I traced the escape-check test by hand through source.py: docs/../../x.md goes through docs -> real to ../x.md, so it trips the check at line 230. With that check removed it would fail later as links to nothing, as the description says. The two halves of KIT_PATCH are sliced cleanly. FAKE_TOOL_USES is read by the fixture in the [name, input] form the new parametrization uses. test_repo_budget.py is still under the 1000-line test cap.

No issues found.

Not verified

I could not run the tests or the mutants. The claim that each new test kills its mutant rests on reading the code and the author's stated runs. I did not read the uptake gating code in repo_run, so the two uptake mutants are unchecked.

[vchamarthi]

**No blockers ** CI at the head is not green (0 failing, 18 pending).

No defects found; the three new tests each reach the check they pin. Traced at c0b5b46: docs/../../x.md with docs -> real normalizes to ../x.md after the hop and raises at the second escape check in GitTree.resolve (_norm keeps .., so the first check cannot catch it); SERVER_ONLY_PATCH and FILE_ONLY_PATCH each make one element of uptake_of empty, so any(uptake_of) in repos/runner.py:231 is the only gate both cases pass through; catalog_competitors builds Candidate(name=s.name, description=s.description), which the tuple assertion now checks.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.78%. Comparing base (92d92d3) to head (c0b5b46).
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #198   +/-   ##
=======================================
  Coverage   91.78%   91.78%           
=======================================
  Files         223      223           
  Lines       27623    27623           
  Branches     4549     4549           
=======================================
  Hits        25354    25354           
- Misses       1606     1607    +1     
+ Partials      663      662    -1     

see 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.