fix(packaging): make shared contracts required peers

[drewstone]

Problem: exact Eval and Interface runtime dependencies force nested copies when an application or Runtime selects a newer compatible shared stack.

Change: release Knowledge 7.2.0 with required Eval >=0.144.10 <0.145.0 and Interface >=0.46.1 <0.47.0 peers. Keep exact 0.144.10 and 0.46.1 development pins. The packed-package verifier now rejects a published runtime dependency, validates both peer ranges, and installs a direct consumer with one physical stack.

Why: Eval and Interface own shared identities and schemas. Consumers must select one cohort. Knowledge must never create a nested older copy.

Proof:

• frozen install passed
• 558 tests passed; 12 skipped
• lint, typecheck, build, and syntax checks passed
• packed direct consumer imported 5 public paths with exactly one Eval 0.144.10, Core 0.5.4, and Interface 0.46.1 copy
• official optimizer packed integration passed with exact Eval and Interface peers
• Runtime will run its packed cohort proof after Knowledge is published

Rollback: retain Knowledge 7.1.3 until the npm artifact is verified.

[tangletools]

✅ Auto-approved drewstone PR — 869fa060

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 · 2026-08-11T03:38:29Z}

[tangletools]

🟢 Value Audit — sound

Verdict	sound
Concerns	0 (none)
Heuristic	0.0s
Duplication	0.0s
Interrogation	38.2s (2 bridge agents)
Total	38.2s

💰 Value — sound

Routine exact-pin alignment release bumping Knowledge 7.1.3→7.1.4 onto Eval 0.144.10; follows the repo's established release cadence and is the correct mechanism for single-copy installs.

• What it does: Bumps @tangle-network/agent-knowledge from 7.1.3 to 7.1.4 and re-pins @tangle-network/agent-eval from exact 0.144.8 to exact 0.144.10 across package.json, pnpm-lock.yaml (new integrity hash sha512-mDUf902...), README install snippet, and adds a CHANGELOG 7.1.4 entry. No source code changes.
• Goals it achieves: Ensures a Runtime consumer that holds Eval 0.144.10 installs exactly one Eval copy — the one that owns the complete-or-unknown usage contract and canonical run records — instead of two copies (0.144.8 from Knowledge + 0.144.10 from Runtime). This avoids duplicate-type hazards and guarantees contract ownership lands on the intended version.
• Assessment: Sound. The exact-pin policy on agent-eval and agent-interface (package.json:78-79) is intentional and uniform, and is the correct lever for the single-copy goal in a pnpm-managed graph. The release is fully self-consistent: lockfile integrity updated, README install line matches the new versions, CHANGELOG entry is accurate. Git log shows this is the fourth Eval-alignment release in two weeks (0.1
• Better / existing approach: none — this is the right approach. Considered a caret range (^0.144.10) to auto-pick patches, but that would defeat the single-copy guarantee the exact pin exists to enforce — pnpm would still resolve a differing exact version if a sibling pinned differently. The 4-file release surface (package.json + lockfile + README + CHANGELOG) is the minimum complete set for a published alignment. Searched gi
• Model: opencode/zai-coding-plan/glm-5.2
• Bridge attempts: 2
• Bridge warning: opencode/kimi-for-coding/k2p7: opencode: opencode error

🎯 Usefulness — sound

Routine exact-pin release bump aligning Knowledge's agent-eval dependency with the newly published 0.144.10, matching the established weekly alignment cadence in this repo.

• Integration: The change edits two lines in package.json: the package version (7.1.3 → 7.1.4) and the exact @tangle-network/agent-eval pin (0.144.8 → 0.144.10). npm confirms 0.144.10 is published. The package is consumed by agent-runtime (per AGENTS.md layering diagram: runtime composes knowledge), and the whole point of the release is so a runtime consumer on Eval 0.144.10 installs a single Eval copy instead o
• Fit with existing patterns: Fits the established pattern exactly. The repo's dependency block uses exact pins (no ^) for all three tangle-network internal packages (agent-eval 0.144.10, agent-interface 0.46.1) — this matches sibling releases 7.0.9 through 7.1.3 verbatim. git log shows 'chore(release): align Knowledge with Eval X' is the repeatable release template (commits 84f6079, 140ab0f, 9fc3c6d). No competing approach; e
• Real-world viability: A version bump of an exact pin cannot introduce behavioral risk beyond what Eval 0.144.10 itself carries; the PR body states 558 tests passed, typecheck/build/lint passed, and the packed package was verified to contain one Eval copy. No new code surface, no new edges. Holds up trivially.
• 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 · 20260811T034135Z}

[tangletools]

✅ No Blockers — 869fa060

Review health 100/100 · Reviewer score 95/100 · Confidence 80/100 · 0 findings (none)

	opencode GLM 5.2	opencode DeepSeek v4 Pro	opencode DeepSeek v4 Flash	aggregate
Readiness	95	95	95	95
Confidence	80	80	80	80
Correctness	95	95	95	95
Security	95	95	95	95
Testing	95	95	95	95
Architecture	95	95	95	95

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

Full multi-shot audit completed 4/4 planned shots over 4 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 4/4 planned shots over 4 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 4/4 planned shots over 4 changed files. Global verifier still owns final merge decision.

No findings.

---

_{tangletools · 2026-08-11T03:43:35Z · trace}

[tangletools]

✅ Clean — 869fa060

Full multi-shot audit completed 4/4 planned shots over 4 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 4/4 planned shots over 4 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 4/4 planned shots over 4 changed files. Global verifier still owns final merge decision.

Full immutable report for this review: trace

Summary comment for this run: full summary

---

_{tangletools · 2026-08-11T03:43:35Z · immutable trace}