fix(mcp): make the docstring budget test environment-independent, trim get_change_risk

[RaghavChamadiya]

main is currently red on test_every_tool_docstring_fits_the_schema_budget, reporting get_change_risk at 1669 chars against a 1600 ceiling.

The test's verdict depended on a transitive dependency

It measured fn.__doc__ straight off the registered tool, and @mcp.tool() does not return the same string on every version of the mcp package:

	mcp 1.26.0 (pinned in uv.lock, so what CI installs)	mcp 1.27.1
fn.__doc__	raw, each line's source indentation attached: 1669	run through inspect.cleandoc: 1576

On a long docstring that is roughly a 90 char swing. The result: the budget passed for anyone whose environment had drifted ahead of the lock, and failed in CI, from byte-identical source. Any local run was systematically more lenient than CI.

The test now dedents before measuring, so it scores the prose an agent actually reads rather than the dependency's normalizing behaviour, and no environment can score more leniently than CI.

Normalizing alone would have hidden a real overage

Under the locked version the schema genuinely was shipping 1669 chars, about 417 tokens against the ~400 the budget exists to enforce. So get_change_risk also loses the sentence describing extensions and exclude_patterns, which restated what its own Args block says two paragraphs later.

Measured from source under both behaviours:

BEFORE (main) raw=1669 FAIL | dedented=1576 pass
AFTER  (fix)  raw=1578 pass | dedented=1489 pass

Every registered tool now fits under either interpretation; get_change_risk is the largest remaining at 1578 raw.

Scope

The docstring crossed the limit in #946, and this is not a regression from the two changes that landed after it. docs/MCP_TOOLS.md does not duplicate the removed sentence, and no other test pins it.

Worth deciding separately: uv.lock pins mcp 1.26.0 while newer environments install 1.27.x. Closing that drift deliberately would prevent the next version-dependent surprise.

Testing

26 tests across the docstring, change-risk, tool-selection, tool-table-drift and prior-fixes suites pass.

[repowise-bot]

✅ Health: 7.5 (unchanged)

📋 At a glance
1 hotspot touched · 1 file with recent fix history. Scoped to packages.

📌 Before you merge

• Run .../mcp/test_change_risk.py, .../mcp/test_prior_fixes.py: they depend on the changed files
• Docs that usually track this code: docs/MCP_TOOLS.md

🗺️ Change map

flowchart LR
  subgraph PR ["Changed in this PR (1 with dependents)"]
    f_packages_server_src_repowise_server_mcp_server_tool_change_risk_py[".../mcp_server/tool_change_risk.py 🔥"]:::changed
  end
  f_packages_server_src_repowise_server_mcp_server___init___py[".../mcp_server/__init__.py"]
  f_packages_server_src_repowise_server_mcp_server_tool_change_risk_py --> f_packages_server_src_repowise_server_mcp_server___init___py
  t_tests_unit_server_mcp_test_change_risk_py(["✅ .../mcp/test_change_risk.py"]):::guard
  t_tests_unit_server_mcp_test_change_risk_py -.-> f_packages_server_src_repowise_server_mcp_server_tool_change_risk_py
  classDef changed fill:#dbeafe,stroke:#1d4ed8,color:#1e3a5f
  classDef warn fill:#fef3c7,stroke:#b45309,color:#78350f
  classDef guard fill:#dcfce7,stroke:#15803d,color:#14532d

Loading

_{Solid arrows: code that imports the changed files (1 direct dependent, from the last indexed snapshot). Dashed: history/tests.}

🩹 Review priority _{(files here with the most recent bug-fix history — defects cluster, so review these first)}

• .../mcp_server/tool_change_risk.py — fixed 2× in the last ~6 months

🔎 More signals (1)

🔥 Hotspot touched (1)

• .../mcp_server/tool_change_risk.py — 4 commits/90d, 3 dependents

---

📊 Full report · ⭐ Star Repowise · 📥 Install bot · Last updated 2026-07-20 08:39 UTC
_{Silence on a single PR with [skip repowise] in the title · Per-repo toggle on repowise.dev/settings?tab=bot}