chore(release): 8.0.6 — require Eval 0.146.0

[drewstone]

Unblocks every consumer that needs @tangle-network/agent-eval/multishot/golden. That subpath is new in Eval 0.146.0, and this package's peer window stopped one version short of it.

The window was derived, not chosen

expectedPeerRange() in scripts/lib/peer-range.mjs computes the range from the dev dependency. For a pre-1.0 dependency it emits >=<version> <major.minor+1.0>, because npm locks a 0.x caret to its minor. Requiring Eval 0.145.21 therefore produced >=0.145.21 <0.146.0 automatically. Requiring 0.146.0 produces >=0.146.0 <0.147.0 by the same rule.

Eval 0.146.0 removes nothing

Measured by diffing the published type surfaces of 0.145.21 and 0.146.0 through the TypeScript checker, across every entry point in the exports map:

measure	0.145.21 → 0.146.0
entry points removed	0
top-level exports removed	0
interface members removed	0
top-level exports added	51

The 20 signature changes are type-precision improvements on previously untyped values — env?: any becoming NodeJS.ProcessEnv, Promise<any> becoming Promise<DatabaseSync | null>.

Proof

pnpm typecheck   clean (src + contracts)
pnpm test        61 files passed | 3 skipped, 604 tests passed | 12 skipped
pnpm build       35 files, 2.65 MB
pnpm verify:package
  Verified @tangle-network/agent-knowledge@8.0.6 with one installed copy each of
  agent-eval 0.146.0, agent-core 0.9.4, and agent-interface 1.0.0:
  clean install, 5 imports, skill, CLI version, and re-pack.

[drewstone]

@tangletools review now

[tangletools]

✅ Auto-approved drewstone PR — f1af5758

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.

This approval is provisional. It rests on the audit running. If the audit cannot run — for example the CLI bridge rejects it — this approval is dismissed rather than left standing, so an unrun check never reads as a passing one.

_{tangletools · auto-approval · reason: drewstone_author · 2026-08-16T22:12:58Z}

[tangletools]

✅ Auto-approved drewstone PR — f1af5758

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.

This approval is provisional. It rests on the audit running. If the audit cannot run — for example the CLI bridge rejects it — this approval is dismissed rather than left standing, so an unrun check never reads as a passing one.

_{tangletools · auto-approval · reason: drewstone_author · 2026-08-16T22:13:01Z}

[tangletools]

✅ Auto-approved drewstone PR — f1af5758

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.

This approval is provisional. It rests on the audit running. If the audit cannot run — for example the CLI bridge rejects it — this approval is dismissed rather than left standing, so an unrun check never reads as a passing one.

_{tangletools · auto-approval · reason: drewstone_author · 2026-08-16T22:13:04Z}

[tangletools]

🟢 Value Audit — sound

Verdict	sound
Coverage	2 of 2 lenses (value, usefulness)
Concerns	0 (none)
Heuristic	0.0s
Duplication	0.0s
Interrogation	178.4s (2 bridge agents)
Total	178.4s

💰 Value — sound

A routine, mechanically-derived peer-window bump that moves the agent-eval cohort to 0.146.0 exactly the way this repo's release machinery prescribes; ship.

• What it does: Bumps the package to 8.0.6 and moves the agent-eval peer dependency from >=0.145.21 <0.146.0 to >=0.146.0 <0.147.0 (package.json:82), with the exact dev pin moved from 0.145.21 to 0.146.0 (package.json:89), plus the matching CHANGELOG entry and pnpm-lock regen (commit f1af575 touches all three, per git show f1af575 --stat). The practical delta: consumers installing this version get an eval p
• Goals it achieves: Reopen the peer window for an additive eval release. Because eval is pre-1.0, the repo's rule derives the window as >=V <0.minor+1.0, so the 0.145.21 pin automatically excluded 0.146.0. The change re-admits current eval by advancing the pin one minor. Secondary goal, achieved implicitly: keep the published cohort guarantee intact — exactly one admitted eval minor, one installed copy — while shif
• Assessment: Good, and squarely in the grain. (1) The range is not hand-written: expectedPeerRange() in scripts/lib/peer-range.mjs:33-35 emits >=0.146.0 <0.147.0 for a pre-1.0 pin — I executed the module and it prints exactly the string at package.json:82. (2) The guard enforces it: scripts/verify-package.mjs:69-74 reads the dev pin, derives the expected range, and fails verify:package on any mismatch, s
• Better / existing approach: none — this is the right approach. I checked the two alternatives against the repo's own machinery: (a) a caret ^0.146.0 is semantically identical for 0.x under npm's rule (peer-range.mjs:5-9 documents this), so the explicit range is just the clearer spelling of the same contract — no improvement available; (b) widening to span both minors (e.g. >=0.145.21 <0.147.0) would avoid forcing a simul
• Model: opencode/zai-coding-plan/glm-5.2
• Bridge attempts: 2
• Bridge warning: opencode/kimi-for-coding/k2p7: opencode: opencode error

🎯 Usefulness — sound

A mechanically derived peer-window bump that unblocks co-install with Eval 0.146.0; every claim checks out against the installed package and the repo's own enforcement script.

• Integration: Fully wired. The window is not hand-picked: expectedPeerRange('0.146.0') returns '>=0.146.0 <0.147.0' (scripts/lib/peer-range.mjs:33-35), byte-identical to package.json:82, and scripts/verify-package.mjs:73,294-297 fails the package check if the declared peer ever diverges from the derivation. The devDep (package.json:89) and pnpm-lock.yaml moved to 0.146.0 in the same commit, and CHANGELOG.md doc
• Fit with existing patterns: Follows the repo's established release-alignment grain exactly. git log shows a chain of identical alignment releases (8.0.5, 662c874; 383ce19 introduced the derivation rule this PR applies; 8ee086c; 2128bb2; fd0eaa7). The pre-1.0 exact-minor window is the deliberate policy documented in scripts/lib/peer-range.mjs:5-9 and enforced by the verify scripts — no competing pattern exists.
• Real-world viability: Independently verified beyond the PR's claims: pnpm install --frozen-lockfile clean; pnpm run typecheck (src + contracts) clean against installed 0.146.0; @tangle-network/agent-eval/multishot/golden importable from the installed 0.146.0 (22 exports); the 0.146.0 exports map retains every subpath this repo imports ('.', ./campaign, ./rl, ./experiment all present). The tsc pass is the decisive check
• 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 · 20260816T222525Z}