fix(mcp): apply default row cap in list_stored_queries markdown fallback

[AnnaSuSu]

Summary

• list_stored_queries's markdown fallback applies no default row cap, while the LanceDB path ten lines above defaults to MAX_ROW_LIMIT. With limit omitted, the fallback returns the entire knowledge/sql corpus.
• Fix: apply the same default in the fallback — a one-line change.

Motivation

Both paths live in the same handler and disagree:

try:
    rows, _total = store.list_queries(
        source=source,
        limit=limit if limit is not None else MAX_ROW_LIMIT,   # default cap
    )
except Exception:
    pairs = load_query_pairs(ctx.project)
    if limit is not None:
        pairs = pairs[:limit]                                  # no default cap

Every other MAX_ROW_LIMIT site guards:

site	guard
_query_with_limit_probe	min(effective_limit, MAX_ROW_LIMIT)
_query_cube_with_limit_probe	min(effective_limit, MAX_ROW_LIMIT)
list_stored_queries (LanceDB)	limit if limit is not None else MAX_ROW_LIMIT
list_stored_queries (fallback)	— none

Three of four obey the cap; the fourth doesn't.

Reachability is wider than "user didn't install the memory extra": the except Exception catches any MemoryStore error, so an installed-and-working setup that hits one transient failure silently switches from capped to uncapped and returns the whole corpus to the MCP client.

Fix

pairs = pairs[: limit if limit is not None else MAX_ROW_LIMIT]

This mirrors the sibling path's semantics exactly (default when omitted, honour an explicit limit). I deliberately did not also align the other asymmetry — the probe helpers hard-cap with min(...) while list_stored_queries treats MAX_ROW_LIMIT as a default an explicit larger limit can exceed. Reconciling those two is a behaviour change to an existing API, not a bug fix, so it is out of scope here.

Verification

• New regression tests in tests/unit/test_mcp_server.py (covered by the dedicated CI mcp tests job):
  • test_list_stored_queries_fallback_applies_default_cap — forces the fallback via a raising store, MAX_ROW_LIMIT monkeypatched to 3, 10 pairs available.
    • Without the change: assert 10 == 3 — RED. With it: PASS.
  • test_list_stored_queries_fallback_honours_explicit_limit — reverse anchor: an explicit limit=2 below the cap must still win.
    • Verified load-bearing: replacing the fix with a blanket pairs[:MAX_ROW_LIMIT] reds it.
• Full mcp tests job vs clean main, verbatim: 14 → 16 passed (+2 = exactly the new tests; 0 new failures). Lint clean (just lint).

License

core/** is Apache-2.0 per the repo LICENSE path map — this contribution is within an Apache-2.0 path.

Summary by CodeRabbit

• Bug Fixes

  • Limited stored query results to the default maximum when no limit is specified.
  • Ensured explicitly requested result limits are consistently honored, including fallback scenarios.

• Tests

  • Added coverage for default and custom result limits when the primary query storage service is unavailable.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: bbb24eb7-1741-4b16-b480-26f269bdb61a

📥 Commits

Reviewing files that changed from the base of the PR and between 243eec8 and 84d34df.

📒 Files selected for processing (2)

• core/wren/src/wren/mcp_server.py
• core/wren/tests/unit/test_mcp_server.py

---

Walkthrough

list_stored_queries now caps markdown fallback results by the default maximum when no limit is supplied and honors explicit limits. Unit tests cover both fallback behaviors.

Changes

Stored query fallback limits

Layer / File(s)	Summary
Fallback limit enforcement and validation core/wren/src/wren/mcp_server.py, core/wren/tests/unit/test_mcp_server.py	Fallback results are capped by MAX_ROW_LIMIT or the caller’s explicit limit, with tests covering both cases.

Estimated code review effort: 2 (Simple) | ~5 minutes

Suggested reviewers: ttw225

Poem

A rabbit found queries hopping wide,
So capped their trail with care and pride.
Three by default, or two on request,
The markdown burrow now works best.
“No endless hops!” the bunny cried.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: applying the default row cap in the markdown fallback for list_stored_queries.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[goldmedal]

Thanks @AnnaSuSu 👍