feat(mcp): surface run_id in run-backed tool results for citation (DRC-3532)#1418
Open
iamcxa wants to merge 1 commit into
Open
feat(mcp): surface run_id in run-backed tool results for citation (DRC-3532)#1418iamcxa wants to merge 1 commit into
iamcxa wants to merge 1 commit into
Conversation
…C-3532)
The cloud-backend MCP tools (row_count_diff, profile_diff, value_diff, query,
query_diff, top_k_diff, histogram_diff) routed through _tool_run_backed returned
only run["result"], dropping run_id. The recce-cloud summary agent therefore
never saw the run_id and could not emit deterministic {{run:<run_id>}} citation
markers, forcing fragile server-side fuzzy prose matching.
Merge run_id into the result dict (additive; existing result fields preserved).
Only added when the response carries a run_id and the result is a dict, so
run-less or non-dict responses are untouched (never synthesize a run_id).
Cross-repo (DRC-3532): the recce-cloud summary agent prompt is updated to emit
the markers; server-side marker replacement already exists in recce_task_func.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Signed-off-by: Kent <iamcxa@gmail.com>
Codecov Report✅ All modified and coverable lines are covered by tests.
... and 3 files with indirect coverage changes 🚀 New features to boost your workflow:
|
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why (cross-repo, DRC-3532)
The recce-cloud summary agent wants to deep-link each factual statement in its PR
summary to the exact run it executed, via deterministic
{{run:<run_id>}}inlinemarkers (server-side marker replacement already exists in recce-cloud-infra's
recce_task_func). But the agent never received the run_id: the cloud-backendMCP tools that back the agent's analysis (row_count_diff, profile_diff,
value_diff, query, query_diff, top_k_diff, histogram_diff) route through
_tool_run_backed, which returned onlyrun["result"]and droppedrun_id.Without run_id the agent cannot cite runs deterministically, forcing fragile
server-side fuzzy prose matching (low/unstable coverage, occasional wrong links).
What
_tool_run_backednow mergesrun_idinto the result dict (additive — existingresult fields preserved). Only added when the response actually carries a run_id
and the result is a dict, so run-less or non-dict responses are untouched (the
agent must never be handed, or synthesize, a run_id it was not given).
Test
tests/test_mcp_cloud_backend.py:40 cloud-backend tests + 123 local mcp_server tests green.
Paired change
recce-cloud-infra: the summary agent prompt emits
{{run:<run_id>}}markersusing this run_id; fuzzy linkify is demoted to a legacy fallback. Tracked under
DRC-3532. Durable structured-citation design is DRC-3634.
🤖 Generated with Claude Code