feat(SUP-1503): add GBrain graph eval coverage

[levineam]

Summary

• Add --compare-graph support to the @jarvos/gbrain retrieval eval harness.
• Let eval questions declare graphSeeds, per-question graphDepth, and graph-specific expected evidence under expected.graph.
• Report a separate gbrain_graph engine so graph sidecar success does not mask direct GBrain search misses.
• Document graph sidecar eval usage and update the public template fixture.

Why

SUP-1499 showed direct GBrain search at 6/8 while QMD was 8/8. SUP-1500/PR #22 adds graph recall because the two GBrain misses are cross-source graph traversal questions. This PR turns that manual sidecar proof into eval coverage.

Private validation

Using the private eval fixture with graph seeds:

• Direct GBrain search: 6/8
• QMD: 8/8
• gbrain_graph: 2/2 on the two cross-source misses

Private report: /Users/andrew/clawd/memory/projects/jarvos-gbrain/graph-eval-baseline-2026-05-10.md

Local validation

• node --check modules/jarvos-gbrain/src/index.js && node --check modules/jarvos-gbrain/scripts/jarvos-gbrain.js
• npm test --prefix modules/jarvos-gbrain
• private eval with --compare-qmd --compare-graph --timeout-ms 15000
• node tests/modules-smoke-test.js
• bash scripts/smoke-test.sh
• npm test
• git diff --check

Stack

Base branch: sup-1500-gbrain-graph-recall / PR #22.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 3f3a0e71-9615-40d4-a219-99a8da2e74b2

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch sup-1503-gbrain-graph-eval

---

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 and usage tips.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ea358a21ec

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Fail graph expectations that have no seeds

When --compare-graph is enabled and a question declares expected.graph/graphExpected but graphSeeds is omitted or misspelled, this guard silently skips adding the gbrain_graph engine. The question can still pass on the direct GBrain result, so the graph-sidecar expectation is never exercised and a misconfigured graph eval appears green; detect a graph expectation with no seeds and mark it failed instead of skipping it.

Useful? React with 👍 / 👎.