feat(bench): add safe benchmark refine attempts

[drewstone]

Summary

• add LOOP_ATTEMPTS to run-benchmarks so each task can run answer -> judge -> safe feedback -> retry loops
• reuse the shared runRefineLoop and stop early on pass
• redact answer/gold/reference fields before feedback reaches retry prompts
• make cli-bridge cells first-class by passing BRIDGE_URL/BRIDGE_BEARER and preserving full bridge model ids

Verification

• pnpm exec tsx --test src/run-benchmarks.test.mts src/benchmarks/rag-benchmarks.test.mts: 8/8 pass
• pnpm exec tsc --noEmit: pass
• pnpm exec tsx --test $(find src -name '*.test.mts' -print | sort): 51/51 pass
• git diff --check: pass
• mergeability: git merge-tree --write-tree origin/main HEAD exit 0
• live router RAGBench official slice: one-shot 0/5 mean 0.168; LOOP_ATTEMPTS=2 0/5 mean 0.218, 0 infra errors
• live bridge fixture smoke: BACKEND=bridge CELLS=opencode/deepseek/deepseek-v4-pro LOOP_ATTEMPTS=2 ran 5 tasks; 3 pass, 1 empty-output error, CRAG fake-company fail

[tangletools]

✅ Auto-approved drewstone PR — 7f09bba8

This PR was opened by the trusted drewstone account.
The full PR reviewer audit still runs separately and will publish findings if it detects issues.

_{tangletools · auto-approval · reason: drewstone_author · 2026-07-07T23:37:14Z}

[tangletools]

⚪ Value Audit — audit-incomplete

Verdict	audit-incomplete
Concerns	0 (none)
Heuristic	0.0s
Duplication	0.1s
Interrogation	8.1s (2 bridge agents)
Total	8.2s

💰 Value — error

value agent produced no parseable value-audit JSON.

• Model: opencode/deepseek/deepseek-v4-pro
• Bridge attempts: 3
• Bridge error: opencode/kimi-for-coding/k2p7: bridge stream ended without value-audit content; opencode/zai-coding-plan/glm-5.2: bridge stream ended without value-audit content; opencode/deepseek/deepseek-v4-pro: bridge stream ended without value-audit content

🎯 Usefulness — error

usefulness agent produced no parseable value-audit JSON.

• Model: opencode/deepseek/deepseek-v4-pro
• Bridge attempts: 3
• Bridge error: opencode/zai-coding-plan/glm-5.2: bridge stream ended without value-audit content; opencode/kimi-for-coding/k2p7: bridge stream ended without value-audit content; opencode/deepseek/deepseek-v4-pro: bridge stream ended without value-audit content

No PR concerns were produced because the value/usefulness agent pass did not complete. Treat this audit as incomplete, not as approval.

---

What this audit checks

It judges the change on its merits — not whether it was tasked out in an issue. Unticketed, fast-moving work is fine; the question is whether the change is good and whether a better or existing approach should be used instead.

Pass	What it asks
Heuristic	Vague title? Whitespace-only or cruft-bearing diff? (content signals only)
Duplication	Do added function/class names already exist elsewhere in the repo?
Value Audit	What does it do? What goal does it achieve? Is it good? Better architecture or already-exists?
Usefulness Audit	Does it integrate and fit? Will it hold up in real use and actually get used?

Findings are concerns, not blocks — the human reviewer decides what to do with them.

_{value-audit · 20260707T233919Z}