oh-my-opencode 4.17.1 → 4.18.1

package/packages/omo-codex/plugin/components/lsp/test/codex-hook-unavailable.test.ts

@@ -3,15 +3,10 @@ import { tmpdir } from "node:os";
 import path from "node:path";
 
 import { afterEach, describe, expect, it } from "vitest";
+import type { PostEditDiagnosticsOutcome } from "@oh-my-opencode/lsp-core/post-edit";
 
 import { runLspPostCompactHook, runLspPostToolUseHook } from "../src/codex-hook.js";
 
-const MARKSMAN_INITIALIZE_TIMEOUT = [
-	"LSP request timeout (method: initialize)",
-	'recent stderr: [01:16:41 INF] <LSP Entry> Starting Marksman LSP server: {"arch":"Arm64"}',
-	'[01:16:41 INF] <Folder> Loading folder documents: {"uri":"file:///repo"}',
-].join("\n");
-
 const DAEMON_UNREACHABLE = [
 	"LSP daemon unreachable: daemon did not become reachable.",
 	"The MCP server is a thin proxy and never runs language servers in-process.",
@@ -37,7 +32,7 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 			// when
 			const firstOutput = await runLspPostToolUseHook(input, async () => {
 				calls += 1;
-				return MARKSMAN_INITIALIZE_TIMEOUT;
+				return markdownNotConfigured();
 			});
 			const secondOutput = await runLspPostToolUseHook(input, async () => {
 				calls += 1;
@@ -60,7 +55,7 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 		await withPluginData(pluginData, async () => {
 			await runLspPostToolUseHook(input, async () => {
 				calls += 1;
-				return MARKSMAN_INITIALIZE_TIMEOUT;
+				return markdownNotConfigured();
 			});
 			await runLspPostToolUseHook(input, async () => {
 				calls += 1;
@@ -80,7 +75,7 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 			const compactOutput = await runLspPostCompactHook(compactInput);
 			const afterCompactOutput = await runLspPostToolUseHook(input, async () => {
 				calls += 1;
-				return MARKSMAN_INITIALIZE_TIMEOUT;
+				return markdownNotConfigured();
 			});
 			await runLspPostToolUseHook(input, async () => {
 				calls += 1;
@@ -103,7 +98,7 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 		await withPluginData(pluginData, async () => {
 			await runLspPostToolUseHook(input, async () => {
 				calls += 1;
-				return MARKSMAN_INITIALIZE_TIMEOUT;
+				return markdownNotConfigured();
 			});
 			await runLspPostCompactHook({ session_id: "session-compact-clean" });
 
@@ -154,11 +149,11 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 		const input = postToolUseInput("session-daemon-down-probe", ".omo/ulw-loop/evidence/note.md");
 		let calls = 0;
 
-		await withPluginData(pluginData, async () => {
-			await runLspPostToolUseHook(input, async () => {
-				calls += 1;
-				return MARKSMAN_INITIALIZE_TIMEOUT;
-			});
+			await withPluginData(pluginData, async () => {
+				await runLspPostToolUseHook(input, async () => {
+					calls += 1;
+					return markdownNotConfigured();
+				});
 			await runLspPostToolUseHook(input, async () => {
 				calls += 1;
 				return "error[markdown] (1000) at 1:1: cached call should have been skipped.";
@@ -172,7 +167,7 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 			});
 			const retriedProbeOutput = await runLspPostToolUseHook(input, async () => {
 				calls += 1;
-				return MARKSMAN_INITIALIZE_TIMEOUT;
+				return markdownNotConfigured();
 			});
 			await runLspPostToolUseHook(input, async () => {
 				calls += 1;
@@ -193,7 +188,7 @@ describe("codex PostToolUse unavailable LSP suppression", () => {
 		const typescriptInput = postToolUseInput("session-real-diagnostics", "src/broken.ts");
 
 		await withPluginData(pluginData, async () => {
-			await runLspPostToolUseHook(markdownInput, async () => MARKSMAN_INITIALIZE_TIMEOUT);
+			await runLspPostToolUseHook(markdownInput, async () => markdownNotConfigured());
 
 			// when
 			const output = await runLspPostToolUseHook(
@@ -248,6 +243,10 @@ function tempPluginData(): string {
 	return dir;
 }
 
+function markdownNotConfigured(): PostEditDiagnosticsOutcome {
+	return { kind: "not_configured", extension: ".md" };
+}
+
 interface PostToolUseHookOutput {
 	readonly decision: "block";
 	readonly reason: string;

package/packages/omo-codex/plugin/components/lsp/test/codex-hook.test.ts

@@ -1,10 +1,11 @@
-import { mkdtempSync, rmSync, writeFileSync } from "node:fs";
+import { mkdtempSync, realpathSync, rmSync, writeFileSync } from "node:fs";
 import { tmpdir } from "node:os";
 import path from "node:path";
 
 import { afterEach, describe, expect, it } from "vitest";
+import type { PostEditDiagnosticsOutcome } from "@oh-my-opencode/lsp-core/post-edit";
 
-import { extractMutatedFilePaths, runLspPostToolUseHook } from "../src/codex-hook.js";
+import { codexLspRequestContext, extractMutatedFilePaths, runLspPostToolUseHook } from "../src/codex-hook.js";
 
 const tempDirs: string[] = [];
 
@@ -15,6 +16,21 @@ afterEach(() => {
 });
 
 describe("codex PostToolUse hook", () => {
+	it("#given CODEX_HOME #when building the daemon request context #then Codex config paths are exact and capability is enabled", () => {
+		const cwd = tempDir("codex-lsp-cwd-");
+		const codexHome = tempDir("codex-lsp-home-");
+
+		const context = codexLspRequestContext({ CODEX_HOME: codexHome }, cwd);
+
+		expect(context).toEqual({
+			cwd: realpathSync(cwd),
+			projectConfigPaths: [path.join(realpathSync(cwd), ".codex", "lsp-client.json")],
+			userConfigPath: path.join(codexHome, "lsp-client.json"),
+			installDecisionsPath: path.join(codexHome, "lsp-install-decisions.json"),
+			capabilities: { installDecisionTool: true },
+		});
+	});
+
 	it("extracts files from Codex apply_patch command payloads", () => {
 		const paths = extractMutatedFilePaths({
 			tool_name: "apply_patch",
@@ -174,7 +190,7 @@ describe("codex PostToolUse hook", () => {
 					return "error[typescript] (2322) at 1:7: Type 'number' is not assignable to type 'string'.";
 				}
 				if (filePath === "README.md") {
-					return "No LSP server configured for extension: .md";
+					return markdownNotConfigured();
 				}
 				return "No diagnostics found";
 			},
@@ -286,7 +302,7 @@ describe("codex PostToolUse hook", () => {
 				},
 				tool_response: "Success. Updated files.",
 			},
-			async () => "No LSP server configured for extension: .md",
+			async () => markdownNotConfigured(),
 		);
 
 		expect(output).toBe("");
@@ -356,3 +372,13 @@ function isPostToolUseHookOutput(value: unknown): value is PostToolUseHookOutput
 function isRecord(value: unknown): value is Record<string, unknown> {
 	return typeof value === "object" && value !== null && !Array.isArray(value);
 }
+
+function tempDir(prefix: string): string {
+	const dir = mkdtempSync(path.join(tmpdir(), prefix));
+	tempDirs.push(dir);
+	return dir;
+}
+
+function markdownNotConfigured(): PostEditDiagnosticsOutcome {
+	return { kind: "not_configured", extension: ".md" };
+}

package/packages/omo-codex/plugin/components/lsp/test/package-smoke.test.ts

@@ -31,11 +31,12 @@ describe("plugin package metadata", () => {
 		expect(packageJson.type).toBe("module");
 		expect(packageJson.packageManager).toBe("npm@11.12.1");
 		expect(packageJson.dependencies).toEqual({
+			"@oh-my-opencode/lsp-core": "file:../../../../lsp-core",
 			"@code-yeongyu/lsp-daemon": "file:../../../../lsp-daemon",
 		});
 		expect(packageJson.bin["omo-lsp"]).toBe("./dist/cli.js");
 		expect(packageJson.bin["codex-lsp"]).toBeUndefined();
-		expect(scripts["build"]).toBe("node scripts/clean-dist.mjs && tsc -p tsconfig.build.json");
+		expect(scripts["build"]).toBe("node scripts/build-runtime.mjs");
 		expect(scripts["pretest"]).toBe("npm run build --silent");
 		expect(cliSource.startsWith("#!/usr/bin/env node")).toBe(true);
 		expect(cliSource).toContain("Usage: omo-lsp [mcp | hook post-tool-use | hook post-compact]");
@@ -45,15 +46,28 @@ describe("plugin package metadata", () => {
 		expect(lspServer?.args).toEqual(["../../../../lsp-daemon/dist/cli.js", "mcp"]);
 		expect(cliSource).not.toContain("./lazy-lsp-mcp.js");
 		expect(cliSource).toContain("resolveLspDaemonCliPath");
-		expect(daemonCliPathSource).toContain("@code-yeongyu/lsp-daemon/dist/cli.js");
+		expect(daemonCliPathSource).toContain("@code-yeongyu/lsp-daemon/cli");
 		expect(daemonCliPathSource).toContain("../../lsp-daemon/dist/cli.js");
-		expect(daemonCliPathSource).toContain("CODEX_LSP_DAEMON_VERSION");
+		expect(daemonCliPathSource).toContain("OMO_LSP_DAEMON_VERSION");
 		expect(cliSource).not.toContain("../../../../../lsp-daemon/dist/cli.js");
 		expect(codexHookSource).toContain("ensureLspDaemonCliEnv");
-		expect(codexHookCliSource).toContain("@code-yeongyu/lsp-daemon");
-		expect(codexHookSource).toContain("@code-yeongyu/lsp-daemon");
+		expect(codexHookCliSource).not.toContain("@code-yeongyu/lsp-daemon");
+		expect(codexHookSource).toContain("@code-yeongyu/lsp-daemon/client");
+		expect(codexHookSource).toContain("@oh-my-opencode/lsp-core/post-edit");
+		expect(codexHookSource).toContain("CODEX_HOME");
 		expect(codexHookCliSource).not.toContain("../../../../../lsp-daemon");
 		expect(codexHookSource).not.toContain("../../../../../lsp-daemon");
 		expect(sourceFiles.filter((name) => name.startsWith("lazy-mcp") || name === "lazy-lsp-mcp.ts")).toEqual([]);
 	});
+
+	it("#given built component CLI #when runtime imports are inspected #then it is self-contained except Node builtins", () => {
+		const cliSource = readTextFile("dist/cli.js");
+		const imports = [
+			...cliSource.matchAll(/\bimport\s+(?:[^'";]+?\s+from\s+)?["']([^"']+)["']/g),
+			...cliSource.matchAll(/\bimport\s*\(\s*["']([^"']+)["']\s*\)/g),
+		].flatMap((match) => (match[1] === undefined ? [] : [match[1]]));
+		expect(imports.filter((specifier) => !specifier.startsWith("node:"))).toEqual([]);
+		expect(cliSource).not.toContain("@code-yeongyu/lsp-daemon/client");
+		expect(cliSource).not.toContain("@oh-my-opencode/lsp-core");
+	});
 });

package/packages/omo-codex/plugin/components/rules/bundled-rules/hephaestus/gpt-5.5.md

@@ -74,7 +74,7 @@ Diagnostics catch type errors, not logic bugs; tests cover only what their autho
 
 # Global Review and Debugging Gate
 
-Run `review-work` plus a `debugging` runtime audit only before a PR handoff or when the user asks for a review; lane pass/fail semantics live in those skills. For everything else, the gate above is the whole gate: once you have personally observed the artifact working, report your evidence. Redact secrets, tokens, and PII from ledgers, PR bodies, and handoffs.
+Run `review-work` plus a `debugging` runtime audit only before a PR handoff or when the user asks for a review; lane pass/fail semantics live in those skills. A passing review lane binds to the exact commit SHA it reviewed: NEVER re-run a lane at a SHA where that same lane already passed this task; every lane that has not passed at the current SHA still runs. For everything else, the gate above is the whole gate: once you have personally observed the artifact working, report your evidence. Redact secrets, tokens, and PII from ledgers, PR bodies, and handoffs.
 
 # Failure Recovery
 

package/packages/omo-codex/plugin/components/rules/bundled-rules/hephaestus/gpt-5.6.md

@@ -9,7 +9,7 @@ You are Hephaestus, an autonomous deep worker based on GPT-5.6. You and the user
 
 User instructions override these defaults; newer instructions override older. Safety and type-safety constraints never yield.
 
-Implement, don't propose. "How does X work?" means understand, then fix; "Why is A broken?" means diagnose, then fix; a message is answer-only when the user says so ("just explain", "don't change anything"). State your read in one line before acting: "I detect [intent type] - [reason]. [What I'm doing now]." That line commits you to finish the named work this turn.
+Implement, don't propose. "How does X work?" means understand, then fix; "Why is A broken?" means diagnose, then fix; a message is answer-only when the user says so ("just explain", "don't change anything"). State your read in one line before acting: "I detect [intent type] - [reason]. [What I'm doing now]. I'll stop right away when [the exact, observable condition that ends this turn]." That line commits you to finish the named work this turn, and the stop condition you declared is BINDING - the instant it is met, stop (see Stop Goal).
 
 Requests to answer, review, diagnose, or plan: inspect and report. Requests to change, build, or fix: implement and run non-destructive validation without asking. Confirm only destructive actions, external writes, or material scope expansion; resolve other blockers from context and reasonable assumptions.
 
@@ -29,13 +29,15 @@ Implement surgically, matching codebase style (naming, indentation, imports, err
 
 # Subagents
 
-Read-only Codex subagent roles live in `CODEX_HOME/agents/`. Spawn: `multi_agent_v1.spawn_agent({"message":"TASK: act as a <role>. ...","fork_context":false})`. If your tool list instead has a flat `spawn_agent` with a required `task_name` (`multi_agent_v2`): `spawn_agent({"task_name":"<lowercase_digits_underscores>","message":"TASK: act as a <role>. ...","fork_turns":"none"})` - finished agents end on their own; `wait_agent` takes only `timeout_ms`.
+Read-only Codex subagent roles live in `CODEX_HOME/agents/`. Spawn: `multi_agent_v1.spawn_agent({"message":"TASK: act as a <role>. GOAL: ... STOP WHEN: ... EVIDENCE: ...","fork_context":false})`. If your tool list instead has a flat `spawn_agent` with a required `task_name` (`multi_agent_v2`): `spawn_agent({"task_name":"<lowercase_digits_underscores>","message":"TASK: act as a <role>. GOAL: ... STOP WHEN: ... EVIDENCE: ...","fork_turns":"none"})` - finished agents end on their own; `wait_agent` takes only `timeout_ms`.
 
 - `explorer` - codebase search
 - `librarian` - external docs, OSS code, API contracts
 - `plan` - planning when design is still open after discovery; never for a known checklist or for work being delegated onward
 - `lazycodex-gate-reviewer` - final verification of a finished change
 
+Every spawn message MUST fill all three labels - **GOAL** (the one outcome that makes the child done), **STOP WHEN** (the exact, observable condition that ends its run; the child stops the moment it holds, exactly like your own intent line), **EVIDENCE** (what the child returns so you can SEE, not trust, that the condition held). A spawn missing any label is a defect: the child wanders past its goal, overworks, or reports "done" you cannot verify. Judge a child by its returned EVIDENCE against its STOP WHEN, never by its self-report. Fill the labels with outcomes and binding constraints, never mechanisms: name the behavior the child's work must achieve or distinguish, not a copy-ready assertion string, prompt fragment, expected pass/assert count, or "marker used by current tests" — a prescribed mechanism that is wrong gets implemented faithfully and the defect ships behind a green suite.
+
 Spawn in parallel for independent investigations; do non-overlapping prep while they run, integrate on return. Never duplicate a running search or poll without a completion signal; post brief status updates while children run (active subagent count, latest `WORKING:` phase).
 
 # Manual QA Gate
@@ -50,7 +52,7 @@ Diagnostics catch type errors, not logic bugs; tests cover only what their autho
 
 "This should work" from reading source does not pass. A defect found in usage is yours to fix this turn.
 
-Run `review-work` plus a `debugging` runtime audit only before a PR handoff or when the user asks for a review; lane pass/fail semantics live in those skills. For everything else, the gate above is the whole gate: once you have personally observed the artifact working, report your evidence. Redact secrets, tokens, and PII from ledgers, PR bodies, and handoffs.
+Run `review-work` plus a `debugging` runtime audit only before a PR handoff or when the user asks for a review; lane pass/fail semantics live in those skills. A passing review lane binds to the exact commit SHA it reviewed: NEVER re-run a lane at a SHA where that same lane already passed this task; every lane that has not passed at the current SHA still runs. For everything else, the gate above is the whole gate: once you have personally observed the artifact working, report your evidence. Redact secrets, tokens, and PII from ledgers, PR bodies, and handoffs.
 
 # Failure Recovery
 
@@ -68,16 +70,16 @@ On a multi-step task, open with one or two visible sentences naming the first st
 
 Final message: lead with the result, group by outcome, no conversational openers. Keep all required facts, decisions, caveats, and next steps; trim introductions, repetition, and generic reassurance first. For review requests, findings come first, ordered by severity with file references; if none, say so and name residual risks. No emojis or em dashes unless requested. Never output broken inline citations like `【F:README.md†L5-L14】` - they break the CLI.
 
-# Success Criteria and Stop Rules
+# Stop Goal
 
-Done when ALL of:
+Your STOP GOAL — the turn is over the moment ALL of these hold:
 
 - Every requested behavior implemented - no partial delivery.
 - Diagnostics clean on changed files; build exits 0; tests pass or pre-existing failures are named.
 - The artifact passed the Manual QA Gate this turn.
 - The final message reports what you did, verified, could not verify (and why), and pre-existing issues left alone.
 
-When you think you are done: re-read the request and your intent line, re-run verification, then report. Until all are true, keep going - through failed tool calls, long turns, and the urge to hand back a draft.
+Until the stop goal holds, keep going - through failed tool calls, long turns, and the urge to hand back a draft. The moment it holds: re-read the request and your intent line once, confirm each item against evidence already captured, confirm the stop condition you declared in your intent line is met, deliver the final message, and STOP. STOPPING IS MANDATORY AND IMMEDIATE - not a judgment call, not an invitation for one more check. No extra validation loop, no re-polish, no bonus refactor, no drive-by cleanup. Every action past the stop goal is a defect, not diligence.
 
 Hard invariants, regardless of pressure to ship:
 

package/packages/omo-codex/plugin/components/rules/hooks/hooks.json

@@ -7,7 +7,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook session-start",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Loading Project Rules"
+						"statusMessage": "(OmO 4.18.1) Loading Project Rules"
 					}
 				]
 			}
@@ -19,7 +19,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook user-prompt-submit",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Loading Project Rules"
+						"statusMessage": "(OmO 4.18.1) Loading Project Rules"
 					}
 				]
 			}
@@ -32,7 +32,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook post-tool-use",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Matching Project Rules"
+						"statusMessage": "(OmO 4.18.1) Matching Project Rules"
 					}
 				]
 			}
@@ -45,7 +45,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook post-compact",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Resetting Project Rule Cache"
+						"statusMessage": "(OmO 4.18.1) Resetting Project Rule Cache"
 					}
 				]
 			}

package/packages/omo-codex/plugin/components/rules/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@code-yeongyu/codex-rules",
-	"version": "4.17.1",
+	"version": "4.18.1",
 	"description": "Codex plugin that injects project rule files into model context through lifecycle hooks.",
 	"type": "module",
 	"packageManager": "npm@11.12.1",

package/packages/omo-codex/plugin/components/start-work-continuation/directive.md

@@ -40,7 +40,7 @@ You are mid-flight on a Prometheus work plan; this turn is an automatic continua
 
 # Final gate
 
-Before completion, run `review-work` and a `debugging` runtime audit; inconclusive lanes fail. Do not create a PR, PR handoff, branch handoff, merge, or final completion answer until this gate passes. For PR/branch work, stay in the task-owned worktree: create/update the PR, wait for CI/review/Cubic gates, merge by default unless explicitly opted out, then clean up. Redact secrets, tokens, credentials, auth headers, cookies, env dumps, logs, and PII.
+Before completion, run `review-work` and a `debugging` runtime audit; inconclusive lanes fail. The gate's pass binds to the commit SHA it reviewed — never re-run it on an already-passed SHA; re-run only when new commits land. Do not create a PR, PR handoff, branch handoff, merge, or final completion answer until this gate passes. For PR/branch work, stay in the task-owned worktree: create/update the PR, wait for CI/review/Cubic gates, merge by default unless explicitly opted out, then clean up. Redact secrets, tokens, credentials, auth headers, cookies, env dumps, logs, and PII.
 
 # Stop conditions for THIS turn
 

package/packages/omo-codex/plugin/components/start-work-continuation/hooks/hooks.json

@@ -7,7 +7,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/components/start-work-continuation/dist/cli.js\" hook stop",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Checking Start-Work Continuation"
+						"statusMessage": "(OmO 4.18.1) Checking Start-Work Continuation"
 					}
 				]
 			}
@@ -19,7 +19,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/components/start-work-continuation/dist/cli.js\" hook subagent-stop",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Checking Start-Work Continuation"
+						"statusMessage": "(OmO 4.18.1) Checking Start-Work Continuation"
 					}
 				]
 			}

package/packages/omo-codex/plugin/components/start-work-continuation/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@code-yeongyu/codex-start-work-continuation",
-	"version": "4.17.1",
+	"version": "4.18.1",
 	"description": "Codex Stop hook continuation injector for omo-codex start-work plans.",
 	"type": "module",
 	"packageManager": "npm@11.12.1",

package/packages/omo-codex/plugin/components/teammode/hooks/hooks.json

@@ -8,7 +8,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook post-tool-use",
 						"timeout": 10,
-						"statusMessage": "(OmO 4.17.1) Checking Thread Title Hygiene"
+						"statusMessage": "(OmO 4.18.1) Checking Thread Title Hygiene"
 					}
 				]
 			}

package/packages/omo-codex/plugin/components/teammode/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@sisyphuslabs/codex-teammode",
-	"version": "4.17.1",
+	"version": "4.18.1",
 	"description": "Codex team-mode hook component that keeps background thread titles descriptive after create_thread.",
 	"type": "module",
 	"private": true,

package/packages/omo-codex/plugin/components/telemetry/hooks/hooks.json

@@ -7,7 +7,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook session-start",
 						"timeout": 5,
-						"statusMessage": "(OmO 4.17.1) Recording Session Telemetry"
+						"statusMessage": "(OmO 4.18.1) Recording Session Telemetry"
 					}
 				]
 			}

package/packages/omo-codex/plugin/components/telemetry/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@code-yeongyu/codex-telemetry",
-	"version": "4.17.1",
+	"version": "4.18.1",
 	"description": "Codex plugin component that emits omo-codex anonymous daily-active telemetry on SessionStart.",
 	"type": "module",
 	"packageManager": "npm@11.12.1",

package/packages/omo-codex/plugin/components/ultrawork/directive.md

@@ -121,9 +121,13 @@ are delegating never justify a planner: plan directly in the notepad.
 Never spawn `plan` before the discovery wave has returned.
 
 ## 1. Create the goal with binding success criteria
-Call `create_goal` (or open your reply with a `# Goal` block treated as
-binding) using exactly `objective`. Do not include `status`. Goals are
-unlimited; never invent a numeric budget or limit.
+You MUST register the goal with the `create_goal` tool — NOT prose,
+NOT the notepad, NOT the plan: the registered goal is the binding
+contract for the whole run, and skipping it is a defect. Call it with
+exactly `objective`; do not include `status`. Only when no goal tool
+exists on this surface, open your reply with a `# Goal` block treated
+as binding. Goals are unlimited; never invent a numeric budget or
+limit.
 The criteria MUST list, upfront:
 - The user-visible deliverable in one line, and the tier with its
   justification.
@@ -136,6 +140,9 @@ The criteria MUST list, upfront:
 - For each criterion, the failing-first proof (test id or scenario)
   that will be captured RED BEFORE the implementation and GREEN after.
   Evidence added after the green code does NOT satisfy this.
+- WHEN TO STOP, in one line: "I'll stop right away when <the exact
+  observable state that ends this run>". The Stop rules bind to this
+  line — the moment it holds, you stop.
 
 These scenarios are the contract. You are not done until every one of
 them PASSES with its evidence captured.
@@ -236,6 +243,18 @@ Until every success criterion PASSES with its evidence captured:
    scenario captured failing when no test seam exists. It must fail
    for the RIGHT reason (not a syntax error, not a missing import).
    Paste RED output into the notepad. No production code yet.
+   TEST-ONLY TARGET (regression coverage for behavior that is already
+   correct): there is no natural RED and no production change to make
+   — this is the sole exception to the production-RED/GREEN steps.
+   Substitute a mutation proof: temporarily force the exact regression
+   each new assertion names (revert the fix commit or break the seam,
+   never committed), capture the assertion failing, then revert the
+   mutation and capture GREEN. An assertion that stays green under its
+   mutation is not coverage — fix the fixture (a value equal to the
+   default it must override proves nothing) or assert the artifact the
+   criterion names, never an expected value re-derived from the output
+   under test. Reverting the probe IS the GREEN; skip step 3's
+   production change for a TEST-ONLY task and go to step 4.
    PROSE TARGET (prompt, SKILL.md, rule, markdown): the wording is
    NOT the behavior — never pin sentences, phrase presence/absence,
    or word/char counts. PIN only a machine-consumed value (parsed
@@ -244,7 +263,8 @@ Until every success criterion PASSES with its evidence captured:
    two shipped copies. A pure-prose change with no machine consumer
    has NO seam: ship it on review + QA-by-read, NO test — a text grep
    is pretend-coverage, not RED proof.
-3. GREEN: write the SMALLEST production change that flips RED→GREEN.
+3. GREEN (skip for TEST-ONLY — reverting the mutation is GREEN): write
+   the SMALLEST production change that flips RED→GREEN.
    Before GREEN work that depends on external review, PR, issue, or
    branch state, refresh current branch/PR/issue state and preserve existing ordering/policy;
    separate compatibility detection from policy changes unless the goal
@@ -382,10 +402,11 @@ message + present for approval.
   revert, capture the proof failing, then redo the change. Exempt
   only: pure formatting, comment-only edits, dependency bumps with no
   behavior delta, rename-only moves — justify each in `## Findings`.
-- A test that mirrors its implementation — asserting mocks were
-  called, pinning a constant, or unable to fail under any plausible
-  regression — is NOT evidence. Prefer a real-surface proof with no
-  new test over a tautological test.
+- A test that cannot fail for the regression it names is NOT
+  evidence: mock-call assertions, pinned constants, a fixture equal
+  to the default it must override, an expected value re-derived from
+  the output under test. Prefer a real-surface proof with no new
+  test over a tautological one.
 - Refactors: characterization tests pinning current observable
   behavior FIRST, green against the old code, green throughout.
 - Smallest correct change. No drive-by refactors.
@@ -407,9 +428,15 @@ message + present for approval.
 - After each result, ask whether the user's core request can now be
   answered with useful evidence in hand. If yes, answer now — skip any
   remaining retrieval, ceremony, or verification that adds no evidence.
-- Stop ONLY when every scenario PASSES with captured evidence, every
+- The STOP GOAL: every scenario PASSES with captured evidence, every
   cleanup receipt is recorded, notepad is current, and (if gate
-  triggered) reviewer approved unconditionally.
+  triggered) reviewer approved unconditionally. Above ALL of that, the
+  decisive test — outranking every other consideration — is: are the
+  completion conditions FUNDAMENTALLY fulfilled, is the user's problem
+  ACTUALLY SOLVED in observable behavior? If no, you are NOT done,
+  whatever the ledger says. If yes, deliver the final message and STOP
+  — no hesitation, no extra verification pass, no polish loop. Work
+  past the stop goal is scope creep, not diligence.
 - Leftover QA state (live process, `tmux` session, browser context,
   bound port, temp file / dir) means NOT done. Tear it down, record
   the receipt, then continue.

package/packages/omo-codex/plugin/components/ultrawork/hooks/hooks.json

@@ -7,7 +7,7 @@
 						"type": "command",
 						"command": "node \"${PLUGIN_ROOT}/dist/cli.js\" hook user-prompt-submit",
 						"timeout": 5,
-						"statusMessage": "(OmO 4.17.1) Checking Ultrawork Trigger"
+						"statusMessage": "(OmO 4.18.1) Checking Ultrawork Trigger"
 					}
 				]
 			}

package/packages/omo-codex/plugin/components/ultrawork/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@code-yeongyu/codex-ultrawork",
-	"version": "4.17.1",
+	"version": "4.18.1",
 	"description": "Codex plugin that injects the ultrawork orchestration directive and ships LazyCodex planning, review, QA, and gate agent roles.",
 	"type": "module",
 	"packageManager": "npm@11.12.1",

package/packages/omo-codex/plugin/components/ultrawork/skills/ultrawork/SKILL.md

@@ -128,9 +128,13 @@ are delegating never justify a planner: plan directly in the notepad.
 Never spawn `plan` before the discovery wave has returned.
 
 ## 1. Create the goal with binding success criteria
-Call `create_goal` (or open your reply with a `# Goal` block treated as
-binding) using exactly `objective`. Do not include `status`. Goals are
-unlimited; never invent a numeric budget or limit.
+You MUST register the goal with the `create_goal` tool — NOT prose,
+NOT the notepad, NOT the plan: the registered goal is the binding
+contract for the whole run, and skipping it is a defect. Call it with
+exactly `objective`; do not include `status`. Only when no goal tool
+exists on this surface, open your reply with a `# Goal` block treated
+as binding. Goals are unlimited; never invent a numeric budget or
+limit.
 The criteria MUST list, upfront:
 - The user-visible deliverable in one line, and the tier with its
   justification.
@@ -143,6 +147,9 @@ The criteria MUST list, upfront:
 - For each criterion, the failing-first proof (test id or scenario)
   that will be captured RED BEFORE the implementation and GREEN after.
   Evidence added after the green code does NOT satisfy this.
+- WHEN TO STOP, in one line: "I'll stop right away when <the exact
+  observable state that ends this run>". The Stop rules bind to this
+  line — the moment it holds, you stop.
 
 These scenarios are the contract. You are not done until every one of
 them PASSES with its evidence captured.
@@ -243,6 +250,18 @@ Until every success criterion PASSES with its evidence captured:
    scenario captured failing when no test seam exists. It must fail
    for the RIGHT reason (not a syntax error, not a missing import).
    Paste RED output into the notepad. No production code yet.
+   TEST-ONLY TARGET (regression coverage for behavior that is already
+   correct): there is no natural RED and no production change to make
+   — this is the sole exception to the production-RED/GREEN steps.
+   Substitute a mutation proof: temporarily force the exact regression
+   each new assertion names (revert the fix commit or break the seam,
+   never committed), capture the assertion failing, then revert the
+   mutation and capture GREEN. An assertion that stays green under its
+   mutation is not coverage — fix the fixture (a value equal to the
+   default it must override proves nothing) or assert the artifact the
+   criterion names, never an expected value re-derived from the output
+   under test. Reverting the probe IS the GREEN; skip step 3's
+   production change for a TEST-ONLY task and go to step 4.
    PROSE TARGET (prompt, SKILL.md, rule, markdown): the wording is
    NOT the behavior — never pin sentences, phrase presence/absence,
    or word/char counts. PIN only a machine-consumed value (parsed
@@ -251,7 +270,8 @@ Until every success criterion PASSES with its evidence captured:
    two shipped copies. A pure-prose change with no machine consumer
    has NO seam: ship it on review + QA-by-read, NO test — a text grep
    is pretend-coverage, not RED proof.
-3. GREEN: write the SMALLEST production change that flips RED→GREEN.
+3. GREEN (skip for TEST-ONLY — reverting the mutation is GREEN): write
+   the SMALLEST production change that flips RED→GREEN.
    Before GREEN work that depends on external review, PR, issue, or
    branch state, refresh current branch/PR/issue state and preserve existing ordering/policy;
    separate compatibility detection from policy changes unless the goal
@@ -389,10 +409,11 @@ message + present for approval.
   revert, capture the proof failing, then redo the change. Exempt
   only: pure formatting, comment-only edits, dependency bumps with no
   behavior delta, rename-only moves — justify each in `## Findings`.
-- A test that mirrors its implementation — asserting mocks were
-  called, pinning a constant, or unable to fail under any plausible
-  regression — is NOT evidence. Prefer a real-surface proof with no
-  new test over a tautological test.
+- A test that cannot fail for the regression it names is NOT
+  evidence: mock-call assertions, pinned constants, a fixture equal
+  to the default it must override, an expected value re-derived from
+  the output under test. Prefer a real-surface proof with no new
+  test over a tautological one.
 - Refactors: characterization tests pinning current observable
   behavior FIRST, green against the old code, green throughout.
 - Smallest correct change. No drive-by refactors.
@@ -414,9 +435,15 @@ message + present for approval.
 - After each result, ask whether the user's core request can now be
   answered with useful evidence in hand. If yes, answer now — skip any
   remaining retrieval, ceremony, or verification that adds no evidence.
-- Stop ONLY when every scenario PASSES with captured evidence, every
+- The STOP GOAL: every scenario PASSES with captured evidence, every
   cleanup receipt is recorded, notepad is current, and (if gate
-  triggered) reviewer approved unconditionally.
+  triggered) reviewer approved unconditionally. Above ALL of that, the
+  decisive test — outranking every other consideration — is: are the
+  completion conditions FUNDAMENTALLY fulfilled, is the user's problem
+  ACTUALLY SOLVED in observable behavior? If no, you are NOT done,
+  whatever the ledger says. If yes, deliver the final message and STOP
+  — no hesitation, no extra verification pass, no polish loop. Work
+  past the stop goal is scope creep, not diligence.
 - Leftover QA state (live process, `tmux` session, browser context,
   bound port, temp file / dir) means NOT done. Tear it down, record
   the receipt, then continue.