oh-my-opencode 5.0.0-beta.6 → 5.0.0-beta.62

package/packages/omo-codex/plugin/components/ulw-loop/src/spawn-guard.ts

@@ -1,85 +1,254 @@
-import { existsSync, readdirSync, readFileSync, statSync, writeFileSync } from "node:fs";
-import { join } from "node:path";
+import { randomBytes } from "node:crypto";
+import { existsSync, mkdirSync, readFileSync, renameSync, statSync, writeFileSync } from "node:fs";
+import { dirname, join } from "node:path";
 
 import type { PreToolUsePayload } from "./codex-hook.js";
 import { parsePreToolUsePayload } from "./codex-hook.js";
 import { isFinalRunCompletionCandidate } from "./goal-status.js";
-import { ulwLoopAttemptEvidenceDir, ulwLoopDir } from "./paths.js";
+import { ulwLoopAttemptEvidenceDir, ulwLoopDir, ulwLoopStateLockPath } from "./paths.js";
+import { spawnRoleDenial } from "./spawn-role-guard.js";
+import { isStateLockTimeout, type StateLockOptions, withStateLockSync } from "./state-lock.js";
+import {
+	GATE_REVIEWER_AGENT_NAMES,
+	REVIEWER_ROLES_BY_SURFACE,
+	resolveToolkitSurface,
+	reviewerRolesFor,
+} from "./surface.js";
 import type { UlwLoopPlan } from "./types.js";
 
 // spawn_agent = v1; collaborationspawn_agent = the delimiter-free flattened v2
 // hook token from codex-rs hook_names.rs; collaboration.spawn_agent = the
 // dotted token observed live in the task-1 probe (hook-tool-tokens.txt).
-const SPAWN_TOOL_TOKENS = new Set(["spawn_agent", "collaborationspawn_agent", "collaboration.spawn_agent"]);
-const DEFAULT_FANOUT_LIMIT = 60;
-const GATE_MESSAGE_PATTERN = /lazycodex-gate-reviewer|final gate review/i;
+const SPAWN_TOOL_TOKENS = new Set([
+	"spawn_agent",
+	"multi_agent_v1.spawn_agent",
+	"collaborationspawn_agent",
+	"collaboration.spawn_agent",
+]);
+export const DEFAULT_FANOUT_LIMIT = 24;
+const DEFAULT_REVIEW_SPAWN_LIMIT = 3;
+const GATE_MESSAGE_PATTERN = /lazycodex-gate-reviewer|omo-senpi-gate-reviewer|final gate review/i;
+const REVIEW_AGENT_TYPES = [
+	...Object.values(REVIEWER_ROLES_BY_SURFACE).map((roles) => roles.gateReview),
+	...Object.values(REVIEWER_ROLES_BY_SURFACE).map((roles) => roles.codeReview),
+	...Object.values(REVIEWER_ROLES_BY_SURFACE).map((roles) => roles.manualQa),
+] as const;
+const REVIEW_AGENT_TYPE_SET = new Set<string>(REVIEW_AGENT_TYPES);
 
-export function applySpawnGuards(payload: PreToolUsePayload): string {
+export interface SpawnGuardOptions {
+	readonly lockTimeoutMs?: number;
+}
+
+export function applySpawnGuards(payload: PreToolUsePayload, options: SpawnGuardOptions = {}): string {
+	if (payload.hook_event_name !== "PreToolUse" || !SPAWN_TOOL_TOKENS.has(payload.tool_name)) return "";
+	if (resolveToolkitSurface() === "lazycodex") {
+		const reason = spawnRoleDenial(payload.tool_input);
+		if (reason !== null) return deny(reason);
+	}
+	return applySpawnBudgetGuards(payload, options);
+}
+
+export function applySpawnBudgetGuards(payload: PreToolUsePayload, options: SpawnGuardOptions = {}): string {
 	if (payload.hook_event_name !== "PreToolUse" || !SPAWN_TOOL_TOKENS.has(payload.tool_name)) return "";
-	const stateDir = ulwLoopDir(payload.cwd, { sessionId: payload.session_id });
+	const breaker = readAdmissionBreaker(payload.session_id);
+	if (breaker !== null)
+		return deny(
+			`Subagent admission failed earlier in this session (${breaker}). Do not spawn more workers or reviewers; report the capacity block and wait for the user.`,
+		);
+	const scope = { sessionId: payload.session_id } as const;
+	const stateDir = ulwLoopDir(payload.cwd, scope);
 	const plan = readPlan(join(stateDir, "goals.json"));
 	if (plan === null) return "";
-	const fanOutDenial = consumeFanOutBudget(stateDir);
-	if (fanOutDenial !== null) return deny(fanOutDenial);
+	const lockOptions: StateLockOptions =
+		options.lockTimeoutMs === undefined ? {} : { timeoutMs: options.lockTimeoutMs };
+	try {
+		return withStateLockSync(
+			ulwLoopStateLockPath(payload.cwd, scope),
+			() => evaluateGuards(payload, plan, stateDir),
+			lockOptions,
+		);
+	} catch (error) {
+		if (isStateLockTimeout(error))
+			return deny(`ulw-loop spawn guard could not take the session state lock: ${error.message}`);
+		throw error;
+	}
+}
+
+function evaluateGuards(payload: PreToolUsePayload, plan: UlwLoopPlan, stateDir: string): string {
+	const fanOutPeek = peekFanOutBudget(stateDir);
+	if (fanOutPeek !== null) return deny(fanOutPeek);
 	const missingArtifact = missingGateArtifact(payload, plan);
 	if (missingArtifact !== null)
-		return deny(`spawn code-review + QA first; gate audits their artifacts: missing ${missingArtifact}`);
+		return deny(`record manual QA first; gate audits its artifacts: missing ${missingArtifact}`);
+	const reviewDenial = consumeReviewSpawnBudget(payload, plan, stateDir);
+	if (reviewDenial !== null) return deny(reviewDenial);
+	const fanOutDenial = consumeFanOutBudget(stateDir);
+	if (fanOutDenial !== null) return deny(fanOutDenial);
 	return "";
 }
 
+export async function runSpawnAdmissionRecorderCli(
+	stdin: NodeJS.ReadableStream,
+	stdout: NodeJS.WritableStream,
+): Promise<void> {
+	const chunks: Buffer[] = [];
+	for await (const chunk of stdin) chunks.push(Buffer.from(chunk));
+	try {
+		const payload = JSON.parse(Buffer.concat(chunks).toString("utf8")) as Record<string, unknown>;
+		const response =
+			typeof payload["tool_response"] === "string"
+				? payload["tool_response"]
+				: JSON.stringify(payload["tool_response"] ?? "");
+		if (!/too many active cells|AgentLimitReached|max_threads|max_concurrent_threads_per_session/i.test(response))
+			return;
+		const dataDir = process.env["PLUGIN_DATA"];
+		if (typeof dataDir !== "string" || typeof payload["session_id"] !== "string") return;
+		const markerDir = join(dataDir, "spawn-breaker");
+		try {
+			mkdirSync(markerDir, { recursive: true });
+			atomicWriteJson(join(markerDir, `${payload["session_id"]}.json`), {
+				reason: response,
+				at: new Date().toISOString(),
+			});
+		} catch (error) {
+			const message = error instanceof Error ? error.message : String(error);
+			process.stderr.write(`[ulw-loop] spawn-guard: could not persist admission failure: ${message}\n`);
+		}
+	} catch {
+		/* malformed hook input is ignored */
+	}
+	void stdout;
+}
+
 export async function runSpawnGuardCli(stdin: NodeJS.ReadableStream, stdout: NodeJS.WritableStream): Promise<void> {
 	try {
 		const chunks: Buffer[] = [];
 		for await (const chunk of stdin) chunks.push(Buffer.from(chunk));
 		const payload = parsePreToolUsePayload(Buffer.concat(chunks).toString("utf8"));
-		if (payload === null) return;
+		if (payload === null) {
+			stdout.write(deny("LazyCodex spawn guard received an invalid hook payload; role routing was not verified."));
+			return;
+		}
 		const output = applySpawnGuards(payload);
 		if (output.length > 0) stdout.write(output);
 	} catch (error) {
-		if (error instanceof Error) return;
+		stdout.write(deny(`LazyCodex spawn guard failed: ${error instanceof Error ? error.message : String(error)}`));
 	}
 }
 
+// Read-only fan-out eligibility check. Returns a denial reason when the next
+// spawn would exceed the limit, without incrementing the counter. Call this
+// before charging any per-reviewer quota so a saturated global cap cannot
+// silently consume reviewer allowances for spawns that will never run.
+function peekFanOutBudget(stateDir: string): string | null {
+	const counterPath = join(stateDir, "spawn-count.json");
+	const count = readCount(counterPath) + 1;
+	const limit = fanOutLimit();
+	if (count <= limit) return null;
+	return `ulw-loop spawn fan-out cap reached (${count}/${limit}). Consolidate work into the agents already running, or raise OMO_SPAWN_FANOUT_LIMIT if this volume is intentional.`;
+}
+
 // Per-session spawn counter; depth/lineage tracking is descoped — this is a
 // total-volume backstop against fan-out explosions, not a recursion tracker.
 function consumeFanOutBudget(stateDir: string): string | null {
 	const counterPath = join(stateDir, "spawn-count.json");
 	const count = readCount(counterPath) + 1;
-	writeFileSync(counterPath, JSON.stringify({ count }));
+	atomicWriteJson(counterPath, { count });
 	const limit = fanOutLimit();
 	if (count <= limit) return null;
 	return `ulw-loop spawn fan-out cap reached (${count}/${limit}). Consolidate work into the agents already running, or raise OMO_SPAWN_FANOUT_LIMIT if this volume is intentional.`;
 }
 
+function consumeReviewSpawnBudget(payload: PreToolUsePayload, plan: UlwLoopPlan, stateDir: string): string | null {
+	const agentType = reviewAgentType(payload.tool_input);
+	if (agentType === null) return null;
+	const goal =
+		plan.goals.find((candidate) => candidate.id === plan.activeGoalId) ??
+		plan.goals.find((candidate) => isFinalRunCompletionCandidate(plan, candidate));
+	if (goal === undefined) return null;
+	const counterPath = join(stateDir, "review-spawn-counts.json");
+	const limit = reviewSpawnLimit();
+	const counts = readCounts(counterPath);
+	const key = `${agentType}:${goal.id}:a${goal.attempt}`;
+	const count = (counts[key] ?? 0) + 1;
+	if (count > limit)
+		return `ulw-loop reviewer no-progress cap reached (${agentType} ${count}/${limit}) for ${goal.id} attempt ${goal.attempt}. Consolidate existing review findings, or checkpoint and start a new attempt after concrete progress.`;
+	counts[key] = count;
+	atomicWriteJson(counterPath, counts);
+	return null;
+}
+
 function missingGateArtifact(payload: PreToolUsePayload, plan: UlwLoopPlan): string | null {
 	if (!isGateReviewerSpawn(payload.tool_input)) return null;
 	const goal = plan.goals.find((candidate) => isFinalRunCompletionCandidate(plan, candidate));
 	if (goal === undefined || goal.status === "complete") return null;
 	if (!goal.successCriteria.every((criterion) => criterion.status === "pass")) return null;
 	const scope = { sessionId: payload.session_id } as const;
+	const requiredArtifacts = [`${goal.id}-manual-qa.md`];
 	if (plan.evidenceLayoutVersion === 2) {
 		const attemptDir = ulwLoopAttemptEvidenceDir(goal.id, goal.attempt, scope);
-		for (const name of [`${goal.id}-code-review.md`, `${goal.id}-manual-qa.md`]) {
+		for (const name of requiredArtifacts) {
 			const relative = `${attemptDir}/${name}`;
 			if (!isNonEmptyFile(join(payload.cwd, relative))) return relative;
 		}
 		return null;
 	}
-	const flatReport = `.omo/evidence/${goal.id}-code-review.md`;
-	if (!isNonEmptyFile(join(payload.cwd, flatReport))) return flatReport;
-	// v1 manual-QA approximation: any other non-empty evidence file counts.
-	if (!hasOtherEvidenceFile(join(payload.cwd, ".omo", "evidence"), `${goal.id}-code-review.md`))
-		return `.omo/evidence/<any manual-QA artifact besides ${goal.id}-code-review.md>`;
-	return null;
+	const manualQa = `.omo/evidence/${goal.id}-manual-qa.md`;
+	return isNonEmptyFile(join(payload.cwd, manualQa)) ? null : manualQa;
 }
 
 function isGateReviewerSpawn(toolInput: unknown): boolean {
-	if (typeof toolInput !== "object" || toolInput === null) return false;
+	const agentType = reviewAgentType(toolInput);
+	return agentType !== null && GATE_REVIEWER_AGENT_NAMES.has(agentType);
+}
+
+function reviewAgentType(toolInput: unknown): string | null {
+	if (typeof toolInput !== "object" || toolInput === null) return null;
 	const record = toolInput as Record<string, unknown>;
 	const agentType = record["agent_type"];
-	if (typeof agentType === "string") return agentType === "lazycodex-gate-reviewer";
+	if (typeof agentType === "string") {
+		// V1: agent_type is present — only reviewer types proceed; any other type is not a review spawn.
+		if (!REVIEW_AGENT_TYPE_SET.has(agentType)) return null;
+		return activeSurfaceReviewerAlias(agentType);
+	}
 	const message = record["message"];
-	return typeof message === "string" && GATE_MESSAGE_PATTERN.test(message);
+	if (typeof message !== "string") return null;
+	const normalizedMessage = message.toLowerCase();
+	// Explicit "act as <role>" assignment takes priority. If the assigned role is not a reviewer,
+	// treat the spawn as non-review so a message that merely mentions a reviewer name does not
+	// accidentally charge that reviewer's quota.
+	const allRoleNames = [...REVIEW_AGENT_TYPES];
+	const explicitAssignment = allRoleNames
+		.map((name) => ({
+			name,
+			index: normalizedMessage.search(new RegExp(`\\bact as (?:an? )?${name}\\b`)),
+		}))
+		.filter(({ index }) => index >= 0)
+		.sort((left, right) => left.index - right.index)[0];
+	if (explicitAssignment !== undefined) return activeSurfaceReviewerAlias(explicitAssignment.name);
+	// Check for an explicit "act as <non-reviewer-role>" assignment. If found, the spawn is not a
+	// review spawn even if the message body mentions a reviewer name.
+	const nonReviewerActAs = /\bact as (?:an? )?\S+/.test(normalizedMessage);
+	if (nonReviewerActAs) return null;
+	const namedReviewer = REVIEW_AGENT_TYPES.find((name) => normalizedMessage.includes(name));
+	if (namedReviewer !== undefined) return activeSurfaceReviewerAlias(namedReviewer);
+	return GATE_MESSAGE_PATTERN.test(message) ? reviewerRolesFor(resolveToolkitSurface()).gateReview : null;
+}
+
+function activeSurfaceReviewerAlias(reviewer: string): string {
+	const activeRoles = reviewerRolesFor(resolveToolkitSurface());
+	for (const roles of Object.values(REVIEWER_ROLES_BY_SURFACE)) {
+		if (reviewer === roles.codeReview) return activeRoles.codeReview;
+		if (reviewer === roles.manualQa) return activeRoles.manualQa;
+		if (reviewer === roles.gateReview) return activeRoles.gateReview;
+	}
+	return reviewer;
+}
+
+function atomicWriteJson(targetPath: string, data: unknown): void {
+	const tmp = join(dirname(targetPath), `.tmp-${randomBytes(6).toString("hex")}`);
+	writeFileSync(tmp, JSON.stringify(data));
+	renameSync(tmp, targetPath);
 }
 
 function deny(reason: string): string {
@@ -93,6 +262,19 @@ function deny(reason: string): string {
 	})}\n`;
 }
 
+function readAdmissionBreaker(sessionId: string): string | null {
+	const dataDir = process.env["PLUGIN_DATA"];
+	if (typeof dataDir !== "string") return null;
+	try {
+		const value = JSON.parse(readFileSync(join(dataDir, "spawn-breaker", `${sessionId}.json`), "utf8")) as {
+			reason?: unknown;
+		};
+		return typeof value.reason === "string" ? value.reason : "capacity limit";
+	} catch {
+		return null;
+	}
+}
+
 function fanOutLimit(): number {
 	const raw = process.env["OMO_SPAWN_FANOUT_LIMIT"];
 	if (raw === undefined) return DEFAULT_FANOUT_LIMIT;
@@ -100,6 +282,13 @@ function fanOutLimit(): number {
 	return Number.isFinite(parsed) && parsed > 0 ? parsed : DEFAULT_FANOUT_LIMIT;
 }
 
+function reviewSpawnLimit(): number {
+	const raw = process.env["OMO_ULW_LOOP_REVIEW_SPAWN_LIMIT"];
+	if (raw === undefined) return DEFAULT_REVIEW_SPAWN_LIMIT;
+	const parsed = Number.parseInt(raw, 10);
+	return Number.isFinite(parsed) && parsed > 0 ? parsed : DEFAULT_REVIEW_SPAWN_LIMIT;
+}
+
 function isNonEmptyFile(path: string): boolean {
 	try {
 		return existsSync(path) && statSync(path).size > 0;
@@ -109,21 +298,27 @@ function isNonEmptyFile(path: string): boolean {
 	}
 }
 
-function hasOtherEvidenceFile(evidenceDir: string, excludedName: string): boolean {
+function readCount(counterPath: string): number {
 	try {
-		return readdirSync(evidenceDir).some((name) => name !== excludedName && isNonEmptyFile(join(evidenceDir, name)));
+		const parsed = JSON.parse(readFileSync(counterPath, "utf8")) as Record<string, unknown>;
+		return typeof parsed["count"] === "number" && parsed["count"] >= 0 ? parsed["count"] : 0;
 	} catch (error) {
-		if (error instanceof Error) return false;
+		if (error instanceof Error) return 0;
 		throw error;
 	}
 }
 
-function readCount(counterPath: string): number {
+function readCounts(counterPath: string): Record<string, number> {
 	try {
-		const parsed = JSON.parse(readFileSync(counterPath, "utf8")) as Record<string, unknown>;
-		return typeof parsed["count"] === "number" && parsed["count"] >= 0 ? parsed["count"] : 0;
+		const parsed: unknown = JSON.parse(readFileSync(counterPath, "utf8"));
+		if (typeof parsed !== "object" || parsed === null || Array.isArray(parsed)) return {};
+		const counts: Record<string, number> = {};
+		for (const [key, value] of Object.entries(parsed)) {
+			if (typeof value === "number" && value >= 0) counts[key] = value;
+		}
+		return counts;
 	} catch (error) {
-		if (error instanceof Error) return 0;
+		if (error instanceof Error) return {};
 		throw error;
 	}
 }

package/packages/omo-codex/plugin/components/ulw-loop/src/spawn-role-guard.ts

@@ -0,0 +1,22 @@
+// Explicit selectors, pinned by a test to the bundled role TOMLs. The generated
+// default is a non-fork fallback, not a selector: a user can opt out of owning it.
+export const LAZYCODEX_SPAWN_ROLES: ReadonlySet<string> = new Set([
+	"explorer",
+	"lazycodex-clone-fidelity-reviewer",
+	"lazycodex-code-reviewer",
+	"lazycodex-gate-reviewer",
+	"lazycodex-qa-executor",
+	"lazycodex-worker-high",
+	"lazycodex-worker-low",
+	"lazycodex-worker-medium",
+	"librarian",
+	"metis",
+	"momus",
+	"plan",
+]);
+
+export function spawnRoleDenial(input: unknown): string | null {
+	const role = typeof input === "object" && input !== null && "agent_type" in input ? input.agent_type : undefined;
+	if (typeof role === "string" && LAZYCODEX_SPAWN_ROLES.has(role)) return null;
+	return `LazyCodex requires an explicit registered agent_type: ${[...LAZYCODEX_SPAWN_ROLES].join(", ")}. Received ${JSON.stringify(role) ?? "no agent_type"}. Use the matching role and fork_turns: "none" (V2) or fork_context: false (V1), unless full history is deliberately required. The hook cannot see the tool schema; if agent_type is unavailable, stop and report incompatible role routing rather than spawning a generic agent. Describing a role in message does not select its TOML.`;
+}

package/packages/omo-codex/plugin/components/ulw-loop/src/state-lock.ts

@@ -0,0 +1,239 @@
+import { randomUUID } from "node:crypto";
+import { closeSync, mkdirSync, openSync, readFileSync, statSync, unlinkSync, writeSync } from "node:fs";
+import { dirname } from "node:path";
+import { setTimeout as sleep } from "node:timers/promises";
+
+import { UlwLoopError } from "./types.js";
+
+// Cross-process exclusive lock for one ulw-loop state directory. Every CLI
+// invocation is its own process, so the in-process promise chain in plan-io
+// serializes nothing across them; this file-level lock is what makes the
+// read-modify-write of goals.json (and the counters next to it) atomic.
+//
+// Protocol: the lock is a file created with O_EXCL whose body records the
+// owner (pid + a per-acquisition token). A waiter reclaims it only when the
+// owner is provably gone (pid dead) or the body never became a record within
+// `staleMs` (a creator that died mid-write); a live owner is never reclaimed by
+// age alone, because overlapping two bodies is exactly the lost update this lock
+// exists to prevent. Waiters back off and fail closed at `timeoutMs`. Release
+// unlinks only a lock that still carries the releaser's own token.
+
+export const ULW_LOOP_LOCK_TIMEOUT_CODE = "ULW_LOOP_LOCK_TIMEOUT";
+
+export interface StateLockOptions {
+	readonly timeoutMs?: number;
+	readonly staleMs?: number;
+}
+
+interface LockRecord {
+	readonly pid: number;
+	readonly createdAt: string;
+	readonly token: string;
+}
+
+interface LockSnapshot {
+	readonly raw: string;
+	readonly record: LockRecord | null;
+	readonly ageMs: number;
+}
+
+type AttemptOutcome = { readonly kind: "acquired"; readonly token: string } | { readonly kind: "retry" | "wait" };
+
+const DEFAULT_TIMEOUT_MS = 10_000;
+const DEFAULT_STALE_MS = 60_000;
+const MIN_DELAY_MS = 5;
+const MAX_DELAY_MS = 100;
+const SLEEP_CELL = new Int32Array(new SharedArrayBuffer(4));
+
+export async function withStateLock<T>(
+	lockPath: string,
+	fn: () => Promise<T>,
+	options: StateLockOptions = {},
+): Promise<T> {
+	const token = await acquireAsync(lockPath, options);
+	try {
+		return await fn();
+	} finally {
+		release(lockPath, token);
+	}
+}
+
+export function withStateLockSync<T>(lockPath: string, fn: () => T, options: StateLockOptions = {}): T {
+	const token = acquireSync(lockPath, options);
+	try {
+		return fn();
+	} finally {
+		release(lockPath, token);
+	}
+}
+
+export function isStateLockTimeout(error: unknown): error is UlwLoopError {
+	return error instanceof UlwLoopError && error.code === ULW_LOOP_LOCK_TIMEOUT_CODE;
+}
+
+async function acquireAsync(lockPath: string, options: StateLockOptions): Promise<string> {
+	const deadline = Date.now() + (options.timeoutMs ?? DEFAULT_TIMEOUT_MS);
+	const staleMs = options.staleMs ?? DEFAULT_STALE_MS;
+	mkdirSync(dirname(lockPath), { recursive: true });
+	for (let attempt = 0; ; ) {
+		const outcome = attemptOnce(lockPath, staleMs);
+		if (outcome.kind === "acquired") return outcome.token;
+		if (outcome.kind === "retry") continue;
+		if (Date.now() >= deadline) throw lockTimeout(lockPath, options.timeoutMs ?? DEFAULT_TIMEOUT_MS);
+		await sleep(backoffMs(attempt));
+		attempt += 1;
+	}
+}
+
+function acquireSync(lockPath: string, options: StateLockOptions): string {
+	const deadline = Date.now() + (options.timeoutMs ?? DEFAULT_TIMEOUT_MS);
+	const staleMs = options.staleMs ?? DEFAULT_STALE_MS;
+	mkdirSync(dirname(lockPath), { recursive: true });
+	for (let attempt = 0; ; ) {
+		const outcome = attemptOnce(lockPath, staleMs);
+		if (outcome.kind === "acquired") return outcome.token;
+		if (outcome.kind === "retry") continue;
+		if (Date.now() >= deadline) throw lockTimeout(lockPath, options.timeoutMs ?? DEFAULT_TIMEOUT_MS);
+		Atomics.wait(SLEEP_CELL, 0, 0, backoffMs(attempt));
+		attempt += 1;
+	}
+}
+
+// EINTR surfaces raw from macOS fs syscalls under some runtimes; it is a retry, never a verdict.
+function attemptOnce(lockPath: string, staleMs: number): AttemptOutcome {
+	try {
+		const token = tryCreate(lockPath);
+		if (token !== null) return { kind: "acquired", token };
+		const snapshot = readSnapshot(lockPath);
+		if (snapshot === null) return { kind: "retry" };
+		if (isStale(snapshot, staleMs) && reclaim(lockPath, snapshot.raw)) return { kind: "retry" };
+		return { kind: "wait" };
+	} catch (error) {
+		if (hasCode(error, "EINTR")) return { kind: "wait" };
+		throw error;
+	}
+}
+
+function tryCreate(lockPath: string): string | null {
+	let fd: number;
+	try {
+		fd = openSync(lockPath, "wx");
+	} catch (error) {
+		if (hasCode(error, "EEXIST")) return null;
+		throw error;
+	}
+	const record: LockRecord = { pid: process.pid, createdAt: new Date().toISOString(), token: randomUUID() };
+	try {
+		writeSync(fd, JSON.stringify(record));
+	} catch (error) {
+		closeSync(fd);
+		// A failed write (e.g. EINTR) must not leave a partial lock the owner never recorded;
+		// otherwise later readers see a young ownerless lock and only age can retire it.
+		try {
+			unlinkSync(lockPath);
+		} catch {
+			// another process already reclaimed the partial file; leave it alone
+		}
+		throw error;
+	}
+	closeSync(fd);
+	return record.token;
+}
+
+function readSnapshot(lockPath: string): LockSnapshot | null {
+	try {
+		const raw = readFileSync(lockPath, "utf8");
+		const ageMs = Date.now() - statSync(lockPath).mtimeMs;
+		return { raw, record: parseRecord(raw), ageMs };
+	} catch (error) {
+		if (hasCode(error, "ENOENT")) return null;
+		throw error;
+	}
+}
+
+function parseRecord(raw: string): LockRecord | null {
+	try {
+		const parsed: unknown = JSON.parse(raw);
+		if (typeof parsed !== "object" || parsed === null) return null;
+		const record = parsed as Record<string, unknown>;
+		const pid = record["pid"];
+		const createdAt = record["createdAt"];
+		const token = record["token"];
+		if (typeof pid !== "number" || !Number.isInteger(pid) || pid <= 0 || typeof createdAt !== "string") return null;
+		if (typeof token !== "string" || token.length === 0) return null;
+		return { pid, createdAt, token };
+	} catch (error) {
+		if (error instanceof SyntaxError) return null;
+		throw error;
+	}
+}
+
+// A body that is not a record is a lock mid-write (or a foreign/legacy file); only
+// age retires it. A parsable record is stale only when its owner is dead: a live
+// owner running past staleMs is slow, not gone, and the waiter fails closed instead.
+function isStale(snapshot: LockSnapshot, staleMs: number): boolean {
+	if (snapshot.record === null) return snapshot.ageMs > staleMs;
+	return !isProcessAlive(snapshot.record.pid);
+}
+
+function isProcessAlive(pid: number): boolean {
+	try {
+		process.kill(pid, 0);
+		return true;
+	} catch (error) {
+		if (hasCode(error, "ESRCH")) return false;
+		if (hasCode(error, "EPERM")) return true;
+		throw error;
+	}
+}
+
+// Re-read right before unlinking so a lock that changed hands since the stale
+// verdict (a sibling waiter reclaimed and re-acquired it) is left alone.
+function reclaim(lockPath: string, expectedRaw: string): boolean {
+	const current = readSnapshot(lockPath);
+	if (current === null) return true;
+	if (current.raw !== expectedRaw) return false;
+	try {
+		unlinkSync(lockPath);
+	} catch (error) {
+		if (!hasCode(error, "ENOENT")) throw error;
+	}
+	return true;
+}
+
+// Only the acquisition that wrote this token may unlink: if the file now carries
+// another token, a waiter has legitimately taken over and its lock must stand.
+function release(lockPath: string, token: string): void {
+	const current = readSnapshot(lockPath);
+	if (current === null || current.record?.token !== token) return;
+	try {
+		unlinkSync(lockPath);
+	} catch (error) {
+		if (!hasCode(error, "ENOENT")) throw error;
+	}
+}
+
+function backoffMs(attempt: number): number {
+	const exponential = Math.min(MAX_DELAY_MS, MIN_DELAY_MS * 2 ** attempt);
+	return exponential + Math.random() * MIN_DELAY_MS;
+}
+
+function lockTimeout(lockPath: string, timeoutMs: number): UlwLoopError {
+	const holder = readSnapshot(lockPath)?.record;
+	const owner = holder === undefined || holder === null ? "another process" : `pid ${holder.pid}`;
+	return new UlwLoopError(
+		`ulw-loop state lock ${lockPath} is held by ${owner} for more than ${timeoutMs}ms; retry once that process finishes, or delete the lock file if that process is gone.`,
+		ULW_LOOP_LOCK_TIMEOUT_CODE,
+		{
+			details: {
+				lockPath,
+				timeoutMs,
+				...(holder === undefined || holder === null ? {} : { holderPid: holder.pid }),
+			},
+		},
+	);
+}
+
+function hasCode(error: unknown, code: string): boolean {
+	return error instanceof Error && "code" in error && error.code === code;
+}

package/packages/omo-codex/plugin/components/ulw-loop/src/status-next-actions.ts

@@ -0,0 +1,36 @@
+import { hasAllCriteriaPass, isFinalRunCompletionCandidate } from "./goal-status.js";
+import { ULW_LOOP_CREATE_GOALS_COMMAND } from "./plan-missing-recovery.js";
+import type { UlwLoopItem, UlwLoopPlan } from "./types.js";
+
+/**
+ * `status --json` is the one call every agent already makes between steps, so it is
+ * also the cheapest place to answer "what now?" without a second round trip.
+ */
+export function statusNextActions(plan: UlwLoopPlan): readonly string[] {
+	const actions: string[] = [];
+	const active = plan.goals.find((goal) => goal.id === plan.activeGoalId);
+	if (plan.goals.length === 0) actions.push(`No goals yet: bootstrap with \`${ULW_LOOP_CREATE_GOALS_COMMAND}\`.`);
+	else if (active !== undefined) actions.push(...activeGoalActions(plan, active));
+	for (const goal of plan.goals.filter((candidate) => candidate.status === "review_blocked"))
+		actions.push(
+			`${goal.id} is review_blocked: capture the reviewer verdict with \`omo-agent-toolkit ulw-loop record-review-blockers --goal-id ${goal.id} --title "<title>" --objective "<objective>" --evidence "<verdict>" --codex-goal-json '<get_goal json>'\`.`,
+		);
+	if (plan.evidenceLayoutVersion !== 2 || active === undefined)
+		actions.push("plan is evidence-layout v1; artifacts go under .omo/evidence/");
+	return actions;
+}
+
+function activeGoalActions(plan: UlwLoopPlan, active: UlwLoopItem): readonly string[] {
+	const unresolved = active.successCriteria.filter((criterion) => criterion.status !== "pass");
+	if (unresolved.length > 0)
+		return [
+			`${active.id} has ${unresolved.length} unresolved criterion(s) (${unresolved.map((criterion) => criterion.id).join(", ")}): record proof with \`omo-agent-toolkit ulw-loop record-evidence --goal-id ${active.id} --criterion-id <id> --status pass --evidence "<observable proof>"\`.`,
+		];
+	if (hasAllCriteriaPass(active) && isFinalRunCompletionCandidate(plan, active))
+		return [
+			`${active.id} passes every criterion and is the final story: update_goal complete, then checkpoint --print-template to build the final quality gate. Use status --json's currentAttemptDir for all quality-gate artifacts.`,
+		];
+	return [
+		`${active.id} passes every criterion: close it with \`omo-agent-toolkit ulw-loop checkpoint --goal-id ${active.id} --status complete --evidence "<proof>" --codex-goal-json '<get_goal json>'\`. Put quality-gate artifacts under the currentAttemptDir shown by status --json.`,
+	];
+}