docs: add eve evals reporter integration (LAM-1855)

[laminar-coding-agent]

Summary

• New evaluations/eve page documenting the LaminarReporter for eve evals: how each eval maps to a Laminar datapoint, gate vs. soft score encoding, registering the reporter in evals.config.ts, and reading/querying results.
• Two screenshots (list + detail) captured from a verified support-triage run against the local prod stack, including a genuine gate failure.
• Changelog entry under June 2026 and the docs.json nav entry (Evaluations group).

Notes

• Page content was re-verified against the shipped LaminarReporter (lmnr-ts feat/lam-1855-eve-laminar-reporter, commit 1f6c6e6): the datapoint input is the eval id, and metadata carries verdict, status, session id, model id, tool-call names, and failed assertions.
• Validated with npx mintlify broken-links (clean); no em dashes or banned marketing words.

Test plan

• Mintlify preview renders the page under Evaluations → eve reporter
• Both images display at readable size inside their <Frame>
• Internal links resolve (comparing-runs, concepts, manual-evaluation, sql-editor, cli, mcp)

---

Note

Low Risk
Documentation-only changes (MDX, nav, changelog); no application or SDK code in this diff.

Overview
Adds documentation for shipping eve eval runs to Laminar via LaminarReporter: mapping (eval → datapoint, gates as gate:<name> scores, soft assertions, executor output), setup in evals/evals.config.ts, running eve eval, and reading/querying results in the UI, SQL, CLI, and MCP. Includes list and detail screenshots and links to comparing runs and related eval docs.

Navigation: new Evaluations → eve reporter page (evaluations/eve) in docs.json.

Changelog: July 2026 entry for the eve reporter feature.

^{Reviewed by Cursor Bugbot for commit 32f99e3. Bugbot is set up for automated code reviews on this repo. Configure here.}

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
laminarai	🟢 Ready	View Preview	Jun 29, 2026, 9:34 AM

💡 Tip: Enable Workflows to automatically generate PRs for you.

[laminar-coding-agent]

Re-verified this page against eve 0.22.6 (it was written against 0.16.2) and a real end-to-end run of the companion SDK PR (lmnr-ai/lmnr-ts#266) against a local Laminar stack. No changes needed:

• evals/evals.config.ts registration with defineEvalConfig({ reporters: [...] }) works verbatim on 0.22.6.
• The per-eval reporters field on defineEval still exists in 0.22.6 types, so the "scope a destination to a single eval" note holds.
• The console-output excerpt matches 0.22.6's console reporter format (verdict glyph, gates n/m, soft name: % entries).
• Gate scores land as binary gate:<name> and soft scores as numeric, exactly as the "How it maps" section describes — confirmed in ClickHouse evaluation_datapoints and in the UI (group progression chart + per-score columns).
• npx mintlify broken-links passes; nav (docs.json) and changelog entries are in place.

[laminar-coding-agent]

Re-verified this page against eve 0.22.6 (latest) and the implementation in lmnr-ai/lmnr-ts#266:

• Registration snippet (evals/evals.config.ts, defineEvalConfig, reporters: [new LaminarReporter(...)]) matches eve 0.22.6 — ran it in a real eve project end-to-end against a local Laminar stack.
• Score encoding described on the page (gate assertions → binary gate:<name>, soft assertions → numeric) matches what actually lands in evaluation_datapoints.
• The SQL example (simpleJSONExtractFloat(scores, 'gate:succeeded')) verified working via the SQL API (the map-subscript alternative fails on the current deployment, so this is the right form).
• The env-var config path (LMNR_BASE_URL for self-hosted) mentioned here had a client-side port-resolution bug — fixed in LAM-1855: add Laminar reporter for eve evals lmnr-ts#266 (commit 70ae486), so the docs are accurate as written.
• docs.json nav entry, changelog entry, and both images present; mintlify broken-links passes.

No changes needed on this branch.