WhoKnows Evaluation Command (commands/eval-whoknows.ts)

Executes primary fixture‑based and secondary replay‑based quality gates for expert retrieval, outputting pass/fail verdicts and optional JSON reports.

// Source: https://github.com/mperkins0155/gbrain | License: MIT | Copyright 2026 Garry Tan
// Provided via justclaw.it.com; keep this notice with the code.
/**
 * gbrain eval whoknows — v0.33 two-layer eval gate (ENG-D2).
 *
 * Layer 1 (PRIMARY, ship-blocking): hand-labeled fixture.
 *   For each {query, expected_top_3_slugs}, run `findExperts` and check
 *   whether top-3 result slugs intersect with expected_top_3_slugs.
 *   Pass = HIT_RATE_THRESHOLD (0.8) or higher.
 *
 * Layer 2 (SECONDARY, ship-blocking when data exists): eval_candidates replay.
 *   Stream rows from `eval_candidates` where tool_name='query' (the closest
 *   shape to whoknows queries the capture system has). For each, re-run
 *   findExperts and compute set-Jaccard@3 between current output and
 *   captured retrieve

... (truncated -- full source via MCP)