[izadoesdev]

Closes two Greptile findings on merged PRs.

• feat(insights): rank optional context when sources exceed the budget #864 (context ranking on analysis replies): the reply tests injected business loaders without rankBusinessContext, so the new ranking path in resume.ts had no coverage. Adds one case to resume-clarification.integration.test.ts. It drives an analysis reply whose business context exceeds the budget through the real ranker with a fake Jev evaluator, then checks four things. The ranked sources (profile, reply, home, exception) reach the analysis and are saved in the observation's contextSnapshot. trackAgentUsage records the typesafe-ai/jev usage (103 tokens) under the reply's org, website and chatId. The reply succeeds. The native hold is confirmed (reserved 1, confirmed 1, released 0).
• fix(insights): keep conflicting measurements private #851 (nested ternaries): the diagnostic tool name and input selection in investigation-flow.test.ts now use plain if/else. Behavior is unchanged.

Verification

• apps/insights: check-types clean, bun run test 764 pass / 0 fail
• INSIGHTS_INTEGRATION_TESTS=true bun test --isolate src/resume-clarification.integration.test.ts: 8 pass / 0 fail
• With the ranking call in resume.ts disabled, the new case fails (the analysis receives decoration in place of exception)
• biome check clean on both files

---

Summary by cubic

Closes two test-coverage gaps found in review of #864 and #851.

• Adds an integration test that runs an oversized analysis reply through the real context ranker with a fake Jev evaluator, verifying the ranked sources, trackAgentUsage recording, and native hold state.
• Replaces nested ternaries in the diagnostic read fixture in investigation-flow.test.ts with plain if/else, without changing behavior.
• Throws when a saved reply has no observation instead of asserting it exists, across both integration tests.

Verification

• check-types clean, 764 unit tests pass, the integration suite passes, and the new case fails when ranking is disabled; biome check clean.

^{Written for commit db0e0b4. Summary will update on new commits.}

[unkey-deploy]

The latest updates on your projects. Learn more about Unkey Deploy

Name	Status	Preview	Inspect	Updated (UTC)
links (preview)	Ready	Visit Preview	Inspect	Sep 27, 2026 8:11pm

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
databuddy-status	Ready	Preview	Sep 27, 2026 8:11pm UTC

2 Skipped Deployments

Project	Deployment	Actions	Updated
dashboard	Skipped		Sep 27, 2026 8:11pm UTC
documentation	Skipped		Sep 27, 2026 8:11pm UTC

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. 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: Repository UI

Review profile: ASSERTIVE

Plan: Advanced

Run ID: 86a52f07-129e-4ee8-af1c-5d39ede53295

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

---

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.}

[greptile-apps]

Confidence Score: 4/5

[Low risk] Test file refactoring and new test coverage.

The test behavior appears sound, but the explicit repository requirements identified above should be satisfied before merging.

Findings

1.  Prohibited unknown type ▶
2.  Prohibited namespace import ▶
3.  Prohibited mock evidence ▶
4.  Prohibited non-null assertions ▶

Summary

This PR adds an integration test for ranking oversized business context during an analysis reply and replaces nested ternaries in an investigation test with if/else branches.

• The new test checks ranked context delivery and persistence, usage attribution, reply completion, and billing settlement.
• The investigation-test refactor preserves its prior tool and input selection.

_{Reviews (1) · Last reviewed commit: "test(insights): replace nested ternaries..."}