chore(release): 0.54.0 — expose recursive agent atom + workspace seam

[drewstone]

Publishes coordinationDriverAgent, serveCoordinationMcp, driverChild (#304) and runInWorkspace (#305) — merged to main without a version bump, so absent from published 0.53.0 (verified: require('@tangle-network/agent-runtime/runtime').coordinationDriverAgent === undefined on npm latest). A consumer is blocked on these. Additive new exports — no breaking change, no fleet bump. dist exposes all four; 945 tests pass.

[tangletools]

✅ Auto-approved PR — 4fffb68e

Blanket team auto-approval is enabled for this reviewer service.
The full PR reviewer audit still runs separately and will publish findings if it detects issues.

_{tangletools · auto-approval · reason: blanket_auto_approve · 2026-06-16T10:56:54Z}

[drewstone]

🟢 Value Audit — sound

Verdict	sound
Concerns	0 (none)
Heuristic	0.0s
Duplication	0.0s
Interrogation	125.3s (2 bridge agents)
Total	125.3s

💰 Value — sound

Bumps package.json from 0.53.0 to 0.54.0 so npm publishes the recursive driver/coordination atom and workspace seam already merged to main; a standard, additive release commit in the repo's established pattern.

• What it does: Changes exactly one line in package.json: the version field moves from "0.53.0" to "0.54.0" (package.json:3). This triggers an npm publish of the current main, which already contains the new exports coordinationDriverAgent, serveCoordinationMcp, driverChild (src/runtime/index.ts:296-316) and runInWorkspace (src/runtime/index.ts:408-413) but was never released because the prior merge landed after 0
• Goals it achieves: Makes the recursive agent-driver primitives (cheap in-process driver, MCP-fronted coordination server, recursive driver-child executor) and the compounding workspace runner available to external consumers via the npm registry. The PR body notes a consumer is currently blocked because require('@tangle-network/agent-runtime/runtime').coordinationDriverAgent is undefined on npm latest; after this bum
• Assessment: Good on its merits. The version bump is the minimal, correct action to release already-shipped code. The bump is a minor 0.x increment (0.53.0 → 0.54.0), appropriate for additive, backward-compatible features. It matches the repo's release convention: the previous release commit (0.53.0, 15ccfad) also changed only package.json, so this is consistent with how the project releases.
• Better / existing approach: none — this is the right approach. The codebase already exposes the four primitives through the ./runtime subpath barrel (src/runtime/index.ts) and tsup builds that barrel into dist/runtime.js (tsup.config.ts:11). A manual package.json version bump is the established release mechanism here (git show 15ccfad), so no materially simpler or more idiomatic alternative exists within this repo's grain.

🎯 Usefulness — sound

The 0.54.0 version bump publishes the recursive coordination driver and workspace seam that are already wired into the runtime entry, exercised by real tests and benchmarks, and unblocks an external consumer.

• Integration: The four new surfaces are exported from the published ./runtime subpath: coordinationDriverAgent at src/runtime/index.ts:298, driverChild at src/runtime/index.ts:308, serveCoordinationMcp at src/runtime/index.ts:316, and runInWorkspace at src/runtime/index.ts:413. The package's exports map ./runtime to dist/runtime.js (package.json:47-51), and tsup.config.ts:11 builds `
• Fit with existing patterns: It extends the existing supervise subsystem (Scope, BudgetPool, executor registry, MCP server) rather than inventing a parallel pattern. runInWorkspace/gitWorkspace/jjWorkspace are a new durable shared-workspace seam, distinct from the in-process fanout helper in src/mcp/worktree.ts (which creates throwaway worktrees for parallel harness runs without committing). The recursive driver
• Real-world viability: The tests go beyond the happy path: coordination-driver.test.ts runs a driver spawning a driver spawning a worker through real scope spawns; coordination-mcp.test.ts sends real HTTP JSON-RPC calls to serveCoordinationMcp; workspace.test.ts:138 verifies typed conflict detection on concurrent commits and workspace.test.ts:79 proves sequential workers compound state across fresh clones. The

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 · 20260616T110046Z}

[tangletools]

✅ No Blockers — 4fffb68e

Readiness 95/100 · Confidence 65/100 · 0 findings (none)

	deepseek	glm	aggregate
Readiness	95	95	95
Confidence	65	65	65
Correctness	95	95	95
Security	95	95	95
Testing	95	95	95
Architecture	95	95	95

Full multi-shot audit completed 1/1 planned shots over 1 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 1/1 planned shots over 1 changed files. Global verifier still owns final merge decision.

No findings.

---

_{tangletools · 2026-06-16T11:00:49Z · trace}

[tangletools]

✅ Auto-approved PR — cd15f1db

Blanket team auto-approval is enabled for this reviewer service.
The full PR reviewer audit still runs separately and will publish findings if it detects issues.

_{tangletools · auto-approval · reason: blanket_auto_approve · 2026-06-16T11:01:54Z}