chore(release): align knowledge with agent-eval 0.131

[drewstone]

Summary

• release @tangle-network/agent-knowledge 6.1.1
• require the published @tangle-network/agent-eval 0.131.0 package
• verify exact Eval and Interface versions in a clean packed install

Checks

• pnpm typecheck
• pnpm test (468 passed, 12 skipped)
• pnpm lint
• pnpm build
• pnpm verify:package
• TMPDIR=/dev/shm pnpm verify:official-optimizers

[tangletools]

✅ Auto-approved drewstone PR — af6c4d13

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 · 2026-07-27T09:55:40Z}

[tangletools]

⚠️ Review Interrupted — af6c4d13

The review runner stopped before publishing a final verdict: no_active_process.

State	Detail
Interrupted	no active process

No review verdict was produced for this run. Trigger a fresh review on the current PR head if the PR is still open.

_{tangletools · #88 · model: kimi-for-coding · updated 2026-07-27T10:30:23Z}

[tangletools]

✅ Auto-approved drewstone PR — af6c4d13

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 · 2026-07-27T10:47:16Z}

[tangletools]

🟢 Value Audit — sound

Verdict	sound
Concerns	0 (none)
Heuristic	0.0s
Duplication	0.0s
Interrogation	78.0s (2 bridge agents)
Total	78.0s

💰 Value — sound

Patch release that bumps agent-eval to 0.131.0 and closes a real verification gap by extending the clean-room package check to also pin and assert the agent-interface version — mirrors the existing agent-eval check exactly.

• What it does: Two changes: (1) package.json bumps @tangle-network/agent-knowledge 6.1.0→6.1.1 and its @tangle-network/agent-eval dependency 0.130.1→0.131.0 (agent-interface stays at 0.35.0). (2) scripts/verify-package.mjs extends the existing agent-eval pin+install+version-assert block to agent-interface: reads the pinned version from package.json, adds it to the npm install args (verify-package.mjs:98), reads
• Goals it achieves: Ship the agent-eval 0.131.0 alignment, and close a real hole: agent-interface was a direct dependency (package.json:79) but the clean-room verify script never explicitly installed or version-checked it. Before this PR, if agent-eval's transitive agent-interface resolution won over the direct pin, verify-package.mjs would have passed silently. Now both @tangle-network/* siblings are pinned, explici
• Assessment: Clean, minimal, in-grain. The agent-eval block is the established template (pin regex check → explicit install → read installed pkg → assert declared==installed==expected), and this PR copy-mirrors it verbatim for agent-interface. No new abstraction, no over-engineering, no behavior change to shipped runtime code. CHANGELOG.md:3-8 accurately describes both deltas. minimumReleaseAgeExclude already
• Better / existing approach: none — this is the right approach. Searched scripts/verify-package.mjs (the only place doing packed-tarball version verification) and scripts/verify-official-optimizers.mjs (different concern — checks optimizer exports, not versions). No existing equivalent that already verifies agent-interface. One could genericize the two sibling checks into a loop over all @tangle-network/* dependencies, but wi
• Model: opencode/zai-coding-plan/glm-5.2
• Bridge attempts: 2
• Bridge warning: opencode/kimi-for-coding/k2p7: bridge stream ended without value-audit content

🎯 Usefulness — sound

Coherent release pin: bumps agent-eval to 0.131.0 and extends the existing tarball-based verify script to also enforce the agent-interface exact version end-to-end.

• Integration: verify-package.mjs is wired into the documented verify:package npm script (package.json:74) and is run for every release. The agent-interface dependency it newly verifies is genuinely load-bearing — 20 source files import its types (src/agent-candidate.ts:4, src/kb-improvement/contracts.ts:6, src/memory/improvement/*.ts).
• Fit with existing patterns: Matches the established release pattern exactly: agent-eval was already pinned+verified the same way at verify-package.mjs:51-54,119-124,131-138; this PR mirrors that block for agent-interface. Prior CHANGELOG entries (5.0.4, 6.0.0, 6.1.0) confirm agent-eval/agent-interface move in lockstep on every release.
• Real-world viability: Uses a real npm pack + npm install --prefer-online from the tarball in a temp dir, then asserts both the manifest dependency string AND the installed node_modules version match. This catches the actual failure modes (manifest drift, registry resolution mismatch) rather than a happy-path mock.
• Model: opencode/zai-coding-plan/glm-5.2
• Bridge attempts: 1

No concerns — sound change, no better or existing approach found. ✅

---

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 · 20260727T104853Z}

[tangletools]

✅ No Blockers — af6c4d13

Review health 100/100 · Reviewer score 92/100 · Confidence 85/100 · 1 finding (1 low)

glm: Correctness 92 · Security 92 · Testing 92 · Architecture 92

Reviewer score is advisory once the run is complete and the verdict has no blockers.

Full multi-shot audit completed 5/5 planned shots over 5 changed files. Global verifier still owns final merge decision.

🟡 LOW agent-interface verification lacks runtime import smoke test that agent-eval has — scripts/verify-package.mjs

Lines 181-184 run await import('@tangle-network/agent-eval/campaign') and assert required exports (gepaOptimizationMethod, skillOptOptimizationMethod) are functions. The new agent-interface verification (lines 55-58, 98, 125-130, 139-147) only checks install + version pinning, with no await import('@tangle-network/agent-interface') smoke test or requiredAgentInterfaceExports assertion. Impact: a broken agent-interface public surface (e.g., missing named export consumed by agent-knowledge) would pass this verify script si

---

_{tangletools · 2026-07-27T10:50:28Z · trace}

[tangletools]

✅ Approved — 1 non-blocking finding — af6c4d13

Full multi-shot audit completed 5/5 planned shots over 5 changed files. Global verifier still owns final merge decision.

Full immutable report for this review: trace

Summary comment for this run: full summary

---

_{tangletools · 2026-07-27T10:50:28Z · immutable trace}