fix: [#881] dbt_pr_review needs no altimate API key — correct stale tool descriptions#882
Conversation
…ool descriptions The native altimate-core engine that backs `dbt_pr_review` (column lineage, track lineage, equivalence, PII, grade) runs fully offline via the bundled `@altimateai/altimate-core` napi binary — there is no `altimate_core.init` and no API-key gate anywhere in that path. Two tool descriptions still carried a stale Python-bridge-era claim that they "Requires `altimate_core.init()` with API key". When the reviewer agent hits a lint-only run it reads those descriptions and wrongly concludes it needs an altimate API key, mis-diagnosing a manifest-resolution problem as an auth one (the symptom reported in #881). - `altimate-core-column-lineage.ts`: drop the false API-key requirement; state it runs offline via the native engine. - `altimate-core-track-lineage.ts`: same correction. - `docs/usage/dbt-pr-review.md`: add a troubleshooting note clarifying that lint-only means the dbt `manifest.json` did not resolve (wrong path, stale manifest, or wrong working directory) — never an API-key/auth failure. The deterministic verdict needs no altimate key; only the optional advisory LLM lane does. - `test/altimate/altimate-core-tool-descriptions.test.ts`: regression guard pinning both descriptions so the false API-key/`altimate_core.init` claim can never silently return. Closes #881 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
There was a problem hiding this comment.
Claude Code Review
This repository is configured for manual code reviews. Comment @claude review to trigger a review and subscribe this PR to future pushes, or @claude review once for a one-time review.
Tip: disable this comment in your organization's Code Review settings.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (4)
📝 WalkthroughWalkthroughThis PR clarifies that altimate-core lineage tools execute fully offline without requiring API keys. Updated tool descriptions, added regression test validation, and documented troubleshooting guidance for users encountering lint-only behavior. ChangesOffline Execution Clarity
Estimated code review effort🎯 2 (Simple) | ⏱️ ~12 minutes Suggested labels
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
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. Comment |
dev-punia-altimate
left a comment
dev-punia-altimate
left a comment
There was a problem hiding this comment.
Multi-Persona Review — Verdict: ship
This PR correctly fixes a misleading error message by removing false API key requirements from offline-native tool descriptions and adding a regression test. All changes are well-documented, tested, and aligned with offline-first architecture principles. No critical or high-severity issues remain.
14/14 agents completed · 177s · 1 findings (0 critical, 0 high, 0 medium)
Low
- [cto] The documentation update in dbt-pr-review.md is excellent but could be more discoverable. The troubleshooting section is buried under 'Options' and may be missed by users encountering the error. →
docs/docs/usage/dbt-pr-review.md:110- 💡 Consider moving the troubleshooting note to a dedicated 'Troubleshooting' section near the top of the page, or linking to it from the 'Common Errors' section if one exists.
Multi-Persona Review · vllm:qwen3-next-80b (waves) + vllm-fallback (synth) ·
| @@ -110,6 +110,24 @@ Options: | |||
| > clearly labeled, never mistaken for a full verdict. Run `dbt compile` first for | |||
There was a problem hiding this comment.
[LOW · cto] The documentation update in dbt-pr-review.md is excellent but could be more discoverable. The troubleshooting section is buried under 'Options' and may be missed by users encountering the error.
💡 Suggestion: Consider moving the troubleshooting note to a dedicated 'Troubleshooting' section near the top of the page, or linking to it from the 'Common Errors' section if one exists.
Confidence: 70/100
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
…re tool descriptions (#884) Follow-up to #882. Two test-only improvements to the regression guard: - Loosen the false-claim regex from `/requires?\b[^.]{0,40}\bapi key\b/` to `/requires?\b.*?\bapi key\b/` so long-gap variants ("requires an API key for authentication") can't slip through. Still does not match the legitimate "no API key required" copy (there "required" follows "api key"). - Add a source sweep over every `altimate-core-*.ts` tool asserting none contains the stale `altimate_core.init` marker — the unambiguous fingerprint of the Python-bridge-era "needs an API key" claim. No native engine tool calls a dispatcher method by that name, so any occurrence is a regression, and scanning the source auto-covers tools added later (the original guard only covered the two known tools). Closes #883 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2 participants
What does this PR do?
Fixes the confusing "needs an altimate AI API key" symptom reported in #881 when running the
dbt_pr_reviewskill.Root cause: The native altimate-core engine that backs
dbt_pr_review(column lineage, track lineage, equivalence, PII, grade) runs fully offline via the bundled@altimateai/altimate-corenapi binary — there is noaltimate_core.initand no API-key gate anywhere in that path (verified innative/altimate-core.ts+native/dispatcher.ts). But two tool descriptions still carried a stale Python-bridge-era claim that they "Requiresaltimate_core.init()with API key". When the reviewer agent hits a lint-only run, it reads those descriptions and wrongly concludes it needs an API key — mis-diagnosing a manifest-resolution problem as an auth one.Changes:
altimate-core-column-lineage.ts/altimate-core-track-lineage.ts: drop the false API-key requirement; state the tools run offline via the native engine (no key/account).docs/usage/dbt-pr-review.md: add a troubleshooting note — lint-only means the dbtmanifest.jsondid not resolve (wrong path, stale manifest, or wrong working directory), never an auth failure. The deterministic verdict needs no altimate key; only the optional advisory LLM lane does.test/altimate/altimate-core-tool-descriptions.test.ts: regression guard pinning both descriptions so the false claim can't silently return.Type of change
Issue for this PR
Closes #881
How did you verify your code works?
bun run typecheck— clean.bun test test/altimate/altimate-core-tool-descriptions.test.ts— 2 pass (also confirmed it would catch the old text via thealtimate_core.initassertion).native/altimate-core.ts:426/:437) callcore.columnLineage/core.trackLineagedirectly; the dispatcher has no auth layer; schema resolution is local-only (Schema.fromJson).prettier --checkclean on changed files; marker check reports no upstream-shared files touched.Checklist
Summary by cubic
Fixes the misleading “needs Altimate API key” message in
dbt_pr_reviewby correcting tool descriptions and clarifying docs. Prevents false auth errors and explains that lint-only mode is a manifest resolution issue, not auth.altimate_core_column_lineageandaltimate_core_track_lineagedescriptions to state they run fully offline via@altimateai/altimate-core(no API key/account required).dbt_pr_reviewdocs: lint-only means the dbtmanifest.jsondidn’t resolve; check path, freshness, and working directory.altimate-core-tool-descriptions.test.tsto guard against reintroducing the false API-key/altimate_core.initclaim.Written for commit a1acf15. Summary will update on new commits.
Summary by CodeRabbit
Documentation
Tests