oh-my-opencode 5.0.0-beta.8 → 5.0.0-beta.81

package/packages/omo-codex/plugin/test/sync-skills-codex-compatibility.test.mjs

@@ -0,0 +1,168 @@
+import assert from "node:assert/strict";
+import { spawnSync } from "node:child_process";
+import { cp, mkdir, mkdtemp, readdir, readFile, rm, symlink } from "node:fs/promises";
+import { tmpdir } from "node:os";
+import { dirname, join, relative, sep } from "node:path";
+import test from "node:test";
+import { fileURLToPath } from "node:url";
+
+import { canonicalUltraworkDirectiveRelativePath } from "../scripts/canonical-ultrawork-directive.mjs";
+
+import {
+	applyCodexSkillOverlays,
+	codexHarnessToolCompatibility,
+	insertCodexCompatibilityGuidance,
+	reviewWorkAnchor,
+	ulwExecuteOriginalCompletion,
+	ulwExecuteOriginalHardRule,
+} from "../scripts/sync-skills.mjs";
+
+const frontmatter = "---\nname: fixture-sentinel\n---\n\n";
+const opencodeExample = "# SENTINEL_SECTION\ntask(SENTINEL_INPUT)\n";
+const pluginRoot = dirname(dirname(fileURLToPath(import.meta.url)));
+const repositoryRoot = join(pluginRoot, "..", "..");
+const opencodeToolPattern = /\b(?:call_omo_agent|background_output|team_[a-z_]+|task)\s*\(/;
+const sharedSkillsRoot = join(repositoryRoot, "shared-skills", "skills");
+
+test("#given sentinel OpenCode tool content #when compatibility guidance is inserted #then the production artifact is placed after frontmatter and before the tool token", () => {
+	const input = `${frontmatter}${opencodeExample}`;
+
+	const actual = insertCodexCompatibilityGuidance(input);
+
+	assert.equal(actual, `${frontmatter}${codexHarnessToolCompatibility}${opencodeExample}`);
+	assert.ok(actual.indexOf(codexHarnessToolCompatibility) < actual.indexOf(opencodeExample));
+});
+
+test("#given shared skill source anchors #when Codex overlays are applied #then exact anchors transform and match the generated artifacts", async () => {
+	const ulwExecuteSource = await readFile(join(sharedSkillsRoot, "ulw-execute", "SKILL.md"), "utf8");
+	const reviewWorkSource = await readFile(join(sharedSkillsRoot, "review-work", "SKILL.md"), "utf8");
+
+	assert.ok(ulwExecuteSource.includes(ulwExecuteOriginalCompletion), "ulw-execute completion anchor drifted from shared source");
+	assert.ok(ulwExecuteSource.includes(ulwExecuteOriginalHardRule), "ulw-execute hard-rule anchor drifted from shared source");
+	assert.ok(reviewWorkSource.includes(reviewWorkAnchor), "review-work anchor drifted from shared source");
+
+	for (const [skillName, source] of [["ulw-execute", ulwExecuteSource], ["review-work", reviewWorkSource]]) {
+		const compatible = insertCodexCompatibilityGuidance(source);
+		const overlaid = applyCodexSkillOverlays(skillName, compatible);
+		assert.notEqual(overlaid, compatible, `${skillName} overlay must change the shared source`);
+		assert.equal(applyCodexSkillOverlays(skillName, overlaid), overlaid, `${skillName} overlay must be idempotent`);
+		assert.equal(await readFile(join(pluginRoot, "skills", skillName, "SKILL.md"), "utf8"), overlaid);
+	}
+});
+
+test("#given already transformed sentinel content #when compatibility guidance is inserted again #then the transform is idempotent", () => {
+	const once = insertCodexCompatibilityGuidance(`${frontmatter}${opencodeExample}`);
+
+	const twice = insertCodexCompatibilityGuidance(once);
+
+	assert.equal(twice, once);
+});
+
+test("#given a stale generated compatibility block #when guidance is inserted #then the production artifact replaces it", () => {
+	const staleSentinel = "STALE_GENERATED_SENTINEL";
+	const staleGuidance = codexHarnessToolCompatibility.replace("multi_agent_v1.spawn_agent", staleSentinel);
+	assert.notEqual(staleGuidance, codexHarnessToolCompatibility);
+
+	const actual = insertCodexCompatibilityGuidance(`${frontmatter}${staleGuidance}${opencodeExample}`);
+
+	assert.equal(actual, `${frontmatter}${codexHarnessToolCompatibility}${opencodeExample}`);
+	assert.equal(actual.includes(staleSentinel), false);
+});
+
+test("#given a custom compatibility block before an OpenCode tool token #when guidance is inserted #then the custom block is preserved", () => {
+	const generatedHeading = codexHarnessToolCompatibility.slice(0, codexHarnessToolCompatibility.indexOf("\n") + 1);
+	const customBlock = `${generatedHeading}\nCUSTOM_BLOCK_SENTINEL\n\n`;
+	const input = `${frontmatter}${customBlock}${opencodeExample}`;
+
+	const actual = insertCodexCompatibilityGuidance(input);
+
+	assert.equal(actual, input);
+});
+
+test("#given an exported template wrapper containing an OpenCode tool token #when guidance is inserted #then wrapper bytes are preserved", () => {
+	const templateWrapper = "export const TEMPLATE_SENTINEL = `task(INPUT_SENTINEL)`;\n";
+
+	const actual = insertCodexCompatibilityGuidance(templateWrapper);
+
+	assert.equal(actual, `${codexHarnessToolCompatibility}${templateWrapper}`);
+});
+
+test("#given real shared skills that need Codex translation #when aggregate skills are synced #then each generated skill injects the production compatibility artifact once before any remaining OpenCode tool token", async () => {
+	const sharedSkillsRoot = join(repositoryRoot, "shared-skills", "skills");
+	const entries = await readdir(sharedSkillsRoot, { withFileTypes: true });
+	const syncScript = await readFile(join(pluginRoot, "scripts", "sync-skills.mjs"), "utf8");
+	const componentSkillNames = new Set(
+		[...syncScript.matchAll(/\[\s*"([^"]+)"\s*,\s*"components\//g)].map((match) => match[1]),
+	);
+	let checked = 0;
+
+	for (const entry of entries) {
+		if (!entry.isDirectory()) continue;
+		if (componentSkillNames.has(entry.name)) continue;
+		const source = await readFile(join(sharedSkillsRoot, entry.name, "SKILL.md"), "utf8");
+		const sourceToolToken = source.match(opencodeToolPattern)?.[0];
+		if (!sourceToolToken) continue;
+		const generated = await readFile(join(pluginRoot, "skills", entry.name, "SKILL.md"), "utf8");
+
+		assert.equal(generated.split(codexHarnessToolCompatibility).length - 1, 1, entry.name);
+		const compatibilityIndex = generated.indexOf(codexHarnessToolCompatibility);
+		assert.ok(compatibilityIndex >= 0, entry.name);
+		const firstGeneratedToolIndex = generated.search(opencodeToolPattern);
+		assert.ok(firstGeneratedToolIndex >= compatibilityIndex, entry.name);
+		assert.ok(firstGeneratedToolIndex < compatibilityIndex + codexHarnessToolCompatibility.length, entry.name);
+		checked += 1;
+	}
+
+	assert.ok(checked > 0, "at least one shared skill must exercise compatibility injection");
+});
+
+test("#given the aggregate sync implementation #when its skill adaptation pipeline is inspected #then it still applies compatibility guidance before overlays", async () => {
+	const script = await readFile(join(pluginRoot, "scripts", "sync-skills.mjs"), "utf8");
+
+	assert.match(
+		script,
+		/applyCodexSkillOverlays\(\s*skillName,\s*insertCodexCompatibilityGuidance\(content,\s*needsSpawnPayloadGuidance\),?\s*\)/,
+	);
+	assert.match(script, /await adaptSkillForCodex\(skillName\)/);
+});
+
+test("#given the flattened plugin cache the Codex installer produces #when sync-skills runs inside it #then it resolves shared skills through the linked package and writes the skill tree", async () => {
+	// given: installMarketplaceLocally copies plugin/ (minus node_modules) to
+	// <CODEX_HOME>/plugins/cache/<marketplace>/omo/<version>, materializes the canonical directive
+	// under that root, links @oh-my-opencode/shared-skills through the rewritten file: dependency,
+	// and only then runs `npm run sync:skills` there. packages/shared-skills is NOT a sibling of
+	// that directory, so a checkout-relative import cannot resolve in this layout.
+	const codexHome = await mkdtemp(join(tmpdir(), "omo-codex-cache-layout-"));
+	const cachedPluginRoot = join(codexHome, "plugins", "cache", "sisyphuslabs", "omo", "0.0.0-cache-layout");
+	await cp(pluginRoot, cachedPluginRoot, {
+		recursive: true,
+		filter: (source) => {
+			const parts = relative(pluginRoot, source).split(sep);
+			return parts[0] !== "skills" && !parts.includes("node_modules") && !parts.includes(".git");
+		},
+	});
+	const directiveTarget = join(cachedPluginRoot, canonicalUltraworkDirectiveRelativePath);
+	await mkdir(dirname(directiveTarget), { recursive: true });
+	await cp(join(repositoryRoot, "..", canonicalUltraworkDirectiveRelativePath), directiveTarget);
+	const linkedSharedSkills = join(cachedPluginRoot, "node_modules", "@oh-my-opencode", "shared-skills");
+	await mkdir(dirname(linkedSharedSkills), { recursive: true });
+	await symlink(join(repositoryRoot, "shared-skills"), linkedSharedSkills, "dir");
+
+	try {
+		// when: the installer runs `npm run sync:skills` here, whose script is `node scripts/sync-skills.mjs`
+		const result = spawnSync(process.execPath, [join(cachedPluginRoot, "scripts", "sync-skills.mjs")], {
+			cwd: cachedPluginRoot,
+			encoding: "utf8",
+		});
+
+		// then
+		assert.equal(result.status, 0, `sync-skills failed in the cache layout:\n${result.stderr}`);
+		const syncedSkills = (await readdir(join(cachedPluginRoot, "skills"), { withFileTypes: true }))
+			.filter((entry) => entry.isDirectory())
+			.map((entry) => entry.name);
+		assert.ok(syncedSkills.includes("ultrawork"), "component skill ultrawork must be synced");
+		assert.ok(syncedSkills.includes("git-master"), "shared skill git-master must be synced from the linked package");
+	} finally {
+		await rm(codexHome, { recursive: true, force: true });
+	}
+});

package/packages/omo-codex/plugin/test/sync-skills-test-support.mjs

@@ -23,16 +23,19 @@ export const expectedSkills = [
 	"remove-ai-slops",
 	"review-work",
 	"rules",
-	"start-work",
 	"teammode",
 	"ultimate-browsing",
 	"ultrawork",
+	"ulw-execute",
 	"ulw-loop",
 	"ulw-plan",
 	"ulw-research",
 	"visual-qa",
 ];
 
+// `ultrawork` is intentionally absent: its skill body is composed by sync-skills.mjs directly from
+// the canonical packages/prompts-core/prompts/ultrawork/codex.md, so it has no component skill
+// source directory to diff against. Its composition is asserted separately in sync-skills.test.mjs.
 export const componentSkillSources = [
 	["comment-checker", "components/comment-checker/skills/comment-checker"],
 	["lcx-contribute-bug-fix", "components/lcx/skills/lcx-contribute-bug-fix"],
@@ -43,10 +46,19 @@ export const componentSkillSources = [
 	["teammode", "components/teammode/skills/teammode"],
 	["ulw-loop", "components/ulw-loop/skills/ulw-loop"],
 	["ulw-plan", "components/ultrawork/skills/ulw-plan"],
-	["ultrawork", "components/ultrawork/skills/ultrawork"],
 ];
 
+export const canonicalUltraworkDirectiveRelativePath = join(
+	"packages",
+	"prompts-core",
+	"prompts",
+	"ultrawork",
+	"codex.md",
+);
+
 const codexCompatibilityEndMarkers = [
+	"Omit optional keys you do not set. Never send `items: []`, `message: \"\"`, `model: \"\"`, `reasoning_effort: \"\"`, or `service_tier: \"\"` — Codex rejects them (`Items can't be empty`, `reasoning_effort must not be empty`).\n\nFor work likely to exceed one wait cycle, require the child to send `WORKING: <task> - <current phase>` before long passes and `BLOCKED: <reason>` only when progress stops. A `multi_agent_v1.wait_agent` timeout only means no new mailbox update arrived; back off between waits (double the timeout up to ~5 minutes) instead of spinning short cycles. Treat a running child as alive. Fallback only when the child is completed without the deliverable, ack-only after followup, explicitly `BLOCKED:`, or no longer running.\n\n",
+	"On the v2 surface `agent_type` may be ABSENT from the spawn schema (verified 2026-07-11: only `fork_turns`/`message`/`task_name`) — when absent, omit it and describe the role inside `message`; installed role TOMLs cannot be selected on that surface. If a code block below conflicts with this section, this section wins. `fork_context` is rejected on `multi_agent_v2` (`fork_context is not supported in MultiAgentV2; use fork_turns instead`).\n\n",
 	"For work likely to exceed one wait cycle, require the child to send `WORKING: <task> - <current phase>` before long passes and `BLOCKED: <reason>` only when progress stops. A `multi_agent_v1.wait_agent` timeout only means no new mailbox update arrived. Treat a running child as alive. Fallback only when the child is completed without the deliverable, ack-only after followup, explicitly `BLOCKED:`, or no longer running.\n\n",
 	"On `multi_agent_v2` sessions the same `agent_type` applies (the OMO installer exposes it) with `fork_turns` instead of `fork_context`. If a code block below conflicts with this section, this section wins.\n\n",
 	"Role-specific behavior must be described in a self-contained `message`. Use `fork_context: false` to start the child with only the initial prompt (no parent history); use `fork_context: true` only when full parent history is truly required. Include any required conversation context, files, diffs, constraints, and requested skill names directly in the spawned agent's `message`. If a code block below conflicts with this section, this section wins.\n\n",
@@ -69,34 +81,52 @@ export function removeCodexCompatibilityGuidance(content) {
 	return `${content.slice(0, start)}${content.slice(end + endMarker.length)}`;
 }
 
-const startWorkOriginalCompletion = `When all top-level checkboxes in \`## TODOs\` and \`## Final Verification Wave\` are complete:
+const ulwExecuteOriginalCompletion = `When all top-level checkboxes in \`## TODOs\` and \`## Final Verification Wave\` are complete:
 
 1. Run the plan's final verification commands.
-2. For PR/branch work, finish the lifecycle from the task-owned worktree: sync \`.omo/\` state back to the main repo, create or update the PR, wait for review/verification gates, merge by default unless explicitly opted out, and remove the worktree only after successful merge or explicit handoff.
+2. For PR/branch work, finish the lifecycle from the last phase's worktree: sync \`.omo/\` state back to the main repo, create or update the PR, wait for review/verification gates, merge by default unless explicitly opted out, and remove the worktree only after successful merge or explicit handoff.
 3. Remove or mark the Boulder work as completed.
 4. Print an \`ORCHESTRATION COMPLETE\` block with the plan path, verification commands, artifacts, and cleanup receipts.`;
 
-const startWorkCodexCompletion = `When all top-level checkboxes in \`## TODOs\` and \`## Final Verification Wave\` are complete:
+const ulwExecuteCodexCompletion = `When all top-level checkboxes in \`## TODOs\` and \`## Final Verification Wave\` are complete:
 
 1. Run the plan's final verification commands.
-2. Complete the **Global Review and Debugging Gate** before any completion claim, PR creation, PR handoff, branch handoff, or merge:
-   - Invoke the \`review-work\` skill with the final diff, changed files, user goal, constraints, run command, and verification evidence. All five review lanes must return PASS. A timeout, missing deliverable, ack-only child, \`BLOCKED:\`, or inconclusive lane is a gate failure, not approval.
-   - Each passing review lane binds to the exact full commit SHA it reviewed. Immediately append a durable record to \`.omo/start-work/ledger.jsonl\` with the lane name, full SHA, PASS verdict, and report artifact/source. Before same-SHA reuse after any continuation or compaction, re-read the ledger record and require the exact lane/SHA pair; memory, chat history, or an unstamped report is not coverage. New commits require fresh applicable lane coverage.
-   - Run a debugging-oriented runtime audit even when the review passes: name at least three plausible failure hypotheses for the changed surface, run the distinguishing checks against the actual artifact, and append a separate durable record with the audit name, exact full SHA, verdict, and evidence artifact/source to \`.omo/start-work/ledger.jsonl\`. Reuse it only after re-reading an exact audit/SHA match.
-   - If any review lane or debugging hypothesis fails, invoke the \`debugging\` skill, confirm root cause with runtime evidence, add the minimal failing test or reproduction, fix it, rerun the affected verification, then rerun the Global Review and Debugging Gate.
-   - Evidence hygiene is mandatory: redact or mask secrets and sensitive user data before writing \`.omo/start-work/ledger.jsonl\`, a PR body, or a handoff. Never include raw tokens, credentials, auth headers, cookies, API keys, env dumps, private logs, or PII; use concise summaries, lengths, hashes, or short non-sensitive prefixes instead.
-   - If the work includes creating, updating, or handing off a PR, refresh \`git status\` and the PR/branch state from the task-owned worktree after the gate, and include only redacted review/debugging evidence in the PR body or handoff.
-3. Finish the PR/branch lifecycle from its task-owned worktree: sync \`.omo/\` state back to the main repo, create or update the PR when requested, wait for CI/review/Cubic gates, merge by default unless explicitly opted out, and remove the worktree only after successful merge or explicit handoff.
-4. Remove or mark the Boulder work as completed.
-5. Print an \`ORCHESTRATION COMPLETE\` block with the plan path, verification commands, Global Review and Debugging Gate verdict, artifacts, and cleanup receipts.`;
-
-const startWorkOriginalHardRule = "- No completion claim while an applicable ultraqa adversarial class was never probed. Each applicable class needs a captured observable result; each skipped class needs a one-line not-applicable reason in the ledger.\n- No PR/branch implementation, review, or merge in the main worktree; use the task-owned git worktree.\n- No unprefixed session ids in Boulder state. Sessions are always recorded as `codex:<session_id>`.";
-
-const startWorkCodexHardRule = "- No completion claim while an applicable ultraqa adversarial class was never probed. Each applicable class needs a captured observable result; each skipped class needs a one-line not-applicable reason in the ledger.\n- No `ORCHESTRATION COMPLETE`, final response, PR creation, PR handoff, or merge before the Global Review and Debugging Gate passes with recorded evidence.\n- No PR/branch implementation or review in the main worktree; create or use a task-owned git worktree first.\n- No unprefixed session ids in Boulder state. Sessions are always recorded as `codex:<session_id>`.";
+2. Record a self-review in the notepad: re-read the diff, run diagnostics, and capture evidence for every acceptance criterion. Run your own manual QA on the real surface.
+3. Only when the user demanded strict, rigorous, or high-accuracy review, spawn ONE \`lazycodex-gate-reviewer\`; otherwise your self-review is the final verification.
+4. Finish the PR/branch lifecycle from its task-owned worktree: sync \`.omo/\` state back to the main repo, create or update the PR, wait for review/verification gates, merge by default unless explicitly opted out, and remove the worktree only after successful merge or explicit handoff.
+5. Remove or mark the Boulder work as completed.
+6. Print an \`ORCHESTRATION COMPLETE\` block with the plan path, verification commands, artifacts, and cleanup receipts.`;
+
+const ulwExecuteOriginalHardRule = `- No production change before a failing-first proof exists (unit test at a seam, otherwise the failing Manual-QA scenario), and no change to existing behavior before a baseline characterization test pins the current behavior and passes on the unchanged code.
+- No \`--dry-run\` as completion evidence.
+- No tests-only completion claim. A Manual-QA artifact is required.
+- **NO DIRECT IMPLEMENTATION BY THE ORCHESTRATOR.** Root NEVER edits product files, writes tests, or runs QA itself — a spawned worker does.
+- No completion claim while an applicable ultraqa adversarial class was never probed. Each applicable class needs a captured observable result; each skipped class needs a one-line not-applicable reason in the ledger.
+- No implementation, review, or merge in the main checkout; every phase works in its task-owned worktree.
+- No unprefixed session ids in Boulder state. Sessions are always recorded as \`codex:<session_id>\`.
+- No stale-memory execution. The plan and ledger are the durable source of truth.`;
+
+const ulwExecuteCodexHardRule = `- No production change before a failing-first proof exists (unit test at a seam, otherwise the failing Manual-QA scenario), and no change to existing behavior before a baseline characterization test pins the current behavior and passes on the unchanged code.
+- No \`--dry-run\` as completion evidence.
+- No tests-only completion claim. A Manual-QA artifact is required.
+- **NO DIRECT IMPLEMENTATION BY THE ORCHESTRATOR.** Root NEVER edits product files, writes tests, or runs QA itself — a spawned worker does.
+- No completion claim while an applicable ultraqa adversarial class was never probed. Each applicable class needs a captured observable result; each skipped class needs a one-line not-applicable reason in the ledger.
+- No implementation, review, or merge in the main checkout; every phase works in a task-owned worktree.
+- No unprefixed session ids in Boulder state. Sessions are always recorded as \`codex:<session_id>\`.
+- No stale-memory execution. The plan and ledger are the durable source of truth.
+- Codex final verification is the exception to the delegated-QA-only rule above: perform your own manual QA on the real surface and record a self-review before completion.`;
 
 const reviewWorkCodexGate = `
-When \`review-work\` is used as a final implementation, PR, or \`$start-work\`
-gate, it is blocking. A timeout, missing deliverable, ack-only response,
+On Codex, use \`review-work\` only when the user asks for a review or demands
+strict, rigorous, or high-accuracy verification, not automatically for a PR or
+completion claim. Your own manual QA and the main session's self-review are
+the default. Spawn ONE \`lazycodex-gate-reviewer\` only for an explicit demand
+for strict review; otherwise run the review checklist inline and record the
+main session's verdict instead of spawning or waiting for a reviewer. These
+Codex rules override the mandatory-reviewer instructions in the shared skill.
+
+When \`review-work\` is used as a final implementation, PR, or \`$ulw-execute\`
+gate, the selected review is blocking. A timeout, missing deliverable, ack-only response,
 explicit \`BLOCKED:\`, or inconclusive lane is not a pass. Treat that lane as
 failed, investigate the underlying uncertainty with the \`debugging\` skill when
 runtime behavior may be wrong, fix with evidence, and rerun the affected lane
@@ -134,10 +164,10 @@ export function removeCodexSkillOverlays(skillName, content) {
 	if (skillName === "ulw-research") {
 		return content.replace(ulwResearchCodexDeliveryGate, ulwResearchOriginalDeliveryGates);
 	}
-	if (skillName === "start-work") {
+	if (skillName === "ulw-execute") {
 		return content
-			.replace(startWorkCodexCompletion, startWorkOriginalCompletion)
-			.replace(startWorkCodexHardRule, startWorkOriginalHardRule);
+			.replace(ulwExecuteCodexCompletion, ulwExecuteOriginalCompletion)
+			.replace(ulwExecuteCodexHardRule, ulwExecuteOriginalHardRule);
 	}
 	if (skillName === "review-work") {
 		return content.replace(reviewWorkCodexGatePattern, "");

package/packages/omo-codex/plugin/test/sync-skills.test.mjs

@@ -5,8 +5,7 @@ import test from "node:test";
 import { fileURLToPath } from "node:url";
 import { sharedSkillsRootPath } from "@oh-my-opencode/shared-skills";
 import {
-	CONTEXT_PRESSURE_SKILL_BUDGET_BYTES,
-	assertPackagedContentMatches,
+	canonicalUltraworkDirectiveRelativePath,
 	componentSkillSources,
 	expectedSkills,
 	listSkillFiles,
@@ -16,7 +15,6 @@ import {
 
 const root = dirname(dirname(fileURLToPath(import.meta.url)));
 const repoRoot = join(root, "..", "..", "..");
-const opencodeOnlyToolPattern = /\b(?:call_omo_agent|background_output|team_[a-z_]+|task)\s*\(/;
 const generatedSkillMetadataFiles = new Set(["agents/openai.yaml"]);
 
 async function readPackagedSkillFile(...segments) {
@@ -96,6 +94,10 @@ test("#given aggregate Codex skills #when source wiring is inspected #then share
 	assert.equal(rootPackageFiles.includes("packages/shared-skills/skills"), true);
 	assert.equal(sharedSkillDependency, "file:../../shared-skills");
 	assert.match(syncScript, /from "@oh-my-opencode\/shared-skills"/);
+	assert.match(syncScript, /from "@oh-my-opencode\/shared-skills\/skill-source-filter"/);
+	// A checkout-relative path dangles once the installer flattens plugin/ into
+	// <CODEX_HOME>/plugins/cache/<marketplace>/omo/<version> (see canonical-ultrawork-directive.mjs).
+	assert.doesNotMatch(syncScript, /from "\.\.\/\.\.\/\.\.\/shared-skills\//);
 	assert.doesNotMatch(syncScript, /shared-skills",\s*"skills"/);
 });
 
@@ -131,7 +133,6 @@ test("#given a shared skill name collides with a Codex component skill #when agg
 	// when / then
 	assert.notEqual(removeCodexCompatibilityGuidance(aggregateSkill), removeCodexCompatibilityGuidance(sharedSkill));
 	assert.equal(removeCodexCompatibilityGuidance(aggregateSkill), removeCodexCompatibilityGuidance(componentSkill));
-	assert.match(aggregateSkill, /multi_agent_v1/);
 });
 
 test("#given shared skill source tests #when aggregate Codex skills are synced #then source tests are not packaged", async () => {
@@ -180,6 +181,18 @@ test("#given component skill sources #when aggregate Codex component skills are
 	}
 });
 
+test("#given the canonical prompts-core directive #when the aggregate ultrawork skill is inspected #then it wraps the canonical bytes in skill frontmatter", async () => {
+	// given
+	const canonical = await readFile(join(repoRoot, canonicalUltraworkDirectiveRelativePath), "utf8");
+
+	// when
+	const skill = await readFile(join(root, "skills", "ultrawork", "SKILL.md"), "utf8");
+
+	// then
+	assert.match(skill, /^---\r?\nname: ultrawork\r?\n/);
+	assert.equal(removeCodexCompatibilityGuidance(skill).endsWith(canonical), true);
+});
+
 test("#given synced ulw-loop skill #when Codex hint metadata is inspected #then ulw-loop surfaces the ulw-loop alias", async () => {
 	// given
 	const skillRoot = join(root, "skills", "ulw-loop");
@@ -191,9 +204,7 @@ test("#given synced ulw-loop skill #when Codex hint metadata is inspected #then
 	// then
 	assert.match(skill, /^---\r?\nname: ulw-loop\r?\n/m);
 	assert.match(interfaceMetadata, /display_name: "\(OmO\) ulw-loop"/);
-	assert.doesNotMatch(interfaceMetadata, /ulw-loop \/ ulw-loop/);
-	assert.match(interfaceMetadata, /short_description: "Goal-like ultrawork loop for systematic decomposition"/);
-	assert.match(interfaceMetadata, /default_prompt: "Use \$ulw-loop/);
+	assert.match(interfaceMetadata, /^\s*default_prompt:\s*".+"$/m);
 });
 
 test("#given synced ulw-loop skill #when Codex hint metadata is inspected #then ulw-loop remains discoverable as an alias", async () => {
@@ -229,92 +240,11 @@ test("#given synced git-master skill #when inspected #then commits and git histo
 
 	// then
 	assert.match(skill, /^---\r?\nname: git-master\r?\n/m);
-	assert.match(skill, /MUST USE whenever a task needs a commit or git-history investigation/);
-	assert.match(skill, /Commit only the user's requested changes/);
-	assert.match(skill, /Choose the Git tool by the question/);
-	assert.match(skill, /git log -S "text"/);
-	assert.match(skill, /git blame -L start,end -- file/);
 	assert.match(interfaceMetadata, /display_name: "\(OmO\) git-master"/);
 	assert.match(interfaceMetadata, /- "git commit"/);
 	assert.match(interfaceMetadata, /- "history search"/);
 });
 
-test("#given synced ulw-loop skill #when worker guidance is inspected #then context-hygiene guidance matches the source", async () => {
-	// given
-	const sourceSkill = await readFile(
-		join(root, "components", "ulw-loop", "skills", "ulw-loop", "references", "full-workflow.md"),
-		"utf8",
-	);
-	const syncedSkill = await readFile(join(root, "skills", "ulw-loop", "SKILL.md"), "utf8");
-	const syncedWorkflow = await readFile(join(root, "skills", "ulw-loop", "references", "full-workflow.md"), "utf8");
-	// ulw-loop is V2-primary (gpt-5.6 sol/terra use the flat `wait_agent`); the `multi_agent_v1.*`
-	// namespace is documented only as the v1 fallback, so the wait_agent refs accept the bare token.
-	const requiredPatterns = [
-		["wait_agent ref", /\bwait_agent\b/],
-		["local spawned-name tracking", /Track spawned agent names locally/],
-		["wait_agent mailbox path", /wait_agent.*mailbox signals/],
-		["progress status contract", /WORKING:/],
-		["long-running plan/reviewer background guidance", /Plan and reviewer agents may run for a long time/],
-		["bounded plan/reviewer polling", /wait_agent.*cycles/],
-		["exponential backoff wait guard", /double the timeout up to ~5 minutes/],
-		["git-master checkpointing", /git-master/],
-		["touched-path commit-style probe", /touched-path commit history/],
-		["verified work-unit commit", /verified work unit/],
-		["observed commit style", /commit in the observed style/],
-	];
-
-	// when / then
-	for (const [label, pattern] of requiredPatterns) {
-		assert.match(sourceSkill, pattern, `source skill missing ${label}`);
-		assert.match(syncedWorkflow, pattern, `synced workflow missing ${label}`);
-	}
-	assert.match(syncedSkill, /references\/full-workflow\.md/);
-	assert.match(syncedSkill, /wait_agent/);
-	assert.match(syncedSkill, /close_agent/);
-});
-
-test("#given packaged start-work skill #when inspected #then no-plan bootstrap and adversarial verification contracts are shipped", async () => {
-	// given
-	const skillFile = await readPackagedSkillFile("start-work", "SKILL.md");
-
-	// when / then
-	assertPackagedContentMatches(skillFile, [
-		["executes Prometheus plan with Boulder state", /Prometheus work plan[\s\S]*Boulder state/],
-		["bootstraps ulw-plan when no selectable plan exists", /no selectable plan[\s\S]*ulw-plan|ulw-plan[\s\S]*no selectable plan/i],
-		["does not execute work without an approved plan", /approved plan[\s\S]*(?:before|prior to)[\s\S]*execution|execution[\s\S]*(?:requires|needs)[\s\S]*approved plan/i],
-		["keeps hook continuation Boulder-only", /Boulder[\s\S]*(?:continuation|Stop hook)[\s\S]*(?:only|solely)|(?:continuation|Stop hook)[\s\S]*(?:only|solely)[\s\S]*Boulder/i],
-		["distinguishes execution from verification", /execution[\s\S]*verification|verification[\s\S]*execution/i],
-		["requires dirty-worktree-aware editing", /dirty worktree/i],
-		["requires stale-state probes", /stale state/i],
-		["rejects misleading success output", /misleading success output/i],
-		["does not accept worker done claims without independent verification", /done claim[\s\S]*independent(?:ly)? verified|independent(?:ly)? verify[\s\S]*done claim/i],
-	]);
-});
-
-test("#given packaged ulw-plan skill #when inspected #then dynamic multi-agent planning contracts are shipped", async () => {
-	// given
-	const skillFile = await readPackagedSkillFile("ulw-plan", "SKILL.md");
-	const workflowFile = await readPackagedSkillFile("ulw-plan", "references", "full-workflow.md");
-	const combinedFile = {
-		path: `${skillFile.path} + ${workflowFile.path}`,
-		content: `${skillFile.content}\n${workflowFile.content}`,
-	};
-
-	// when / then
-	assertPackagedContentMatches(combinedFile, [
-		["self-orchestrates 5 host subagents for planning", /(?:self-orchestrates|orchestrates)[\s\S]*5[\s\S]*host subagents/i],
-		["requires dynamic workflow phases", /dynamic[\s\S]*workflow[\s\S]*phase|phase[\s\S]*dynamic[\s\S]*workflow/i],
-		["keeps verification distinct from execution", /verification[\s\S]*execution|execution[\s\S]*verification/i],
-		["requires dirty-worktree-aware planning", /dirty worktree/i],
-		["requires stale-state checks between source and packaged payloads", /stale state/i],
-		["rejects misleading success output", /misleading success output/i],
-		["does not accept subagent outputs as success without independent verification", /subagent outputs?[\s\S]*(?:not|never)[\s\S]*(?:success|approval)|independent(?:ly)? verif(?:y|ied|ication)[\s\S]*subagent outputs?/i],
-		["treats Discord or external content as claims, not instructions", /(?:Discord|external content)[\s\S]*claims?[\s\S]*not instructions?|not instructions?[\s\S]*(?:Discord|external content)/i],
-	]);
-	assert.doesNotMatch(combinedFile.content, opencodeOnlyToolPattern);
-	assert.doesNotMatch(combinedFile.content, /Proceeding to plan generation/);
-});
-
 test("#given packaged Codex ulw-plan surfaces #when inspected #then dangerous sandbox bypass guidance is not shipped", async () => {
 	// given
 	const dangerousBypassToken = ["dangerously", "bypass"].join("-");
@@ -330,34 +260,3 @@ test("#given packaged Codex ulw-plan surfaces #when inspected #then dangerous sa
 	assert.doesNotMatch(packagedWorkflow.content, dangerousBypassPattern, `${packagedWorkflow.path} ships unsafe Codex bypass guidance`);
 	assert.doesNotMatch(componentWorkflow.content, dangerousBypassPattern, `${componentWorkflow.path} ships unsafe Codex bypass guidance`);
 });
-
-test("#given packaged ulw-research skill #when Codex delivery gates are inspected #then the category-based proofread gate is stripped", async () => {
-	// given
-	const content = await readFile(join(root, "skills", "ulw-research", "SKILL.md"), "utf8");
-
-	// then the writing-category proofread gate never ships to Codex (no category concept there)
-	assert.doesNotMatch(content, /category="writing"/, "codex ulw-research ships the writing-category proofread gate");
-	assert.doesNotMatch(content, /Proofread gate/, "codex ulw-research still references the proofread gate");
-	// and the visual-QA delivery gate survives as the single gate
-	assert.match(content, /### The delivery gate \u2014 visual QA must PASS/u, "codex ulw-research lost the visual-QA delivery gate");
-	assert.match(content, /Visual QA \(always\)/, "codex ulw-research lost the visual-QA gate body");
-});
-
-test("#given context-pressure-prone skills #when bundled for Codex #then the eagerly loaded payload stays budgeted", async () => {
-	// given
-	const skillsRoot = join(root, "skills");
-	const skillNames = ["debugging", "ulw-loop"];
-
-	// when
-	let totalBytes = 0;
-	for (const skillName of skillNames) {
-		const content = await readFile(join(skillsRoot, skillName, "SKILL.md"), "utf8");
-		totalBytes += Buffer.byteLength(content, "utf8");
-	}
-
-	// then
-	assert.ok(
-		totalBytes <= CONTEXT_PRESSURE_SKILL_BUDGET_BYTES,
-		`debugging + ulw-loop eager payload is ${totalBytes} bytes, above ${CONTEXT_PRESSURE_SKILL_BUDGET_BYTES}`,
-	);
-});

package/packages/omo-codex/plugin/test/teammode-archive-ambiguity.test.mjs

@@ -1,48 +1,8 @@
 import assert from "node:assert/strict";
-import { readFileSync } from "node:fs";
-import { join } from "node:path";
 import test from "node:test";
 
-import { root } from "./aggregate-plugin-fixture.mjs";
 import { cleanupTeamRoot, createTeamRoot, readTeamJson, runTeam } from "./teammode-safety-fixture.mjs";
 
-const teammodeSkillPath = join("components", "teammode", "skills", "teammode", "SKILL.md");
-
-test("#given Codex App thread ids can be ambiguous across hosts #when archive guidance is read #then it separates app-thread archival from team-state archival", () => {
-	const body = readFileSync(join(root, teammodeSkillPath), "utf8");
-
-	assert.match(
-		body,
-		/Ambiguous Codex thread id|ambiguous across hosts/i,
-		"archive guidance must name the Codex App multi-host ambiguity failure instead of only saying the archive tool is unavailable",
-	);
-	assert.match(
-		body,
-		/app-thread archival blocker/i,
-		"archive guidance must classify ambiguous app-thread ids as app-thread archival blockers",
-	);
-	assert.match(
-		body,
-		/record .*team log/i,
-		"archive guidance must require a durable team-log record when app-thread archival cannot be proven",
-	);
-	assert.match(
-		body,
-		/continue .*team-state archive/i,
-		"archive guidance must let the team-state archive proceed after recording the app-thread archival blocker",
-	);
-	assert.match(
-		body,
-		/archive --note/i,
-		"archive guidance must route ambiguous app-thread archive blockers into archive --note evidence",
-	);
-	assert.match(
-		body,
-		/Do not delete/i,
-		"archive guidance must keep evidence available until ambiguous app-thread archival is resolved",
-	);
-});
-
 test("#given app-thread archival is blocked by an ambiguous id #when archive runs with a note #then team state is archived and keeps the blocker in the log", () => {
 	const tempRoot = createTeamRoot("omo-codex-teammode-archive-ambiguity-");
 	try {