@ngockhoale/ukit 2.7.1 → 2.7.2

package/src/cli/adapters.js

@@ -19,16 +19,6 @@ export const OPTIONAL_ADAPTERS = [
       '.codex/agents',
     ],
   },
-  {
-    key: 'opencode',
-    label: 'OpenCode',
-    itemIds: [
-      'multi-opencode-config',
-    ],
-    managedPaths: [
-      'opencode.json',
-    ],
-  },
   {
     key: 'omp',
     label: 'Oh My Pi (omp)',
@@ -75,13 +65,11 @@ export const TOOL_ALIASES = new Map([
   ['claude-code', 'claude'],
   ['codex', 'codex'],
   ['openai-codex', 'codex'],
-  ['opencode', 'opencode'],
-  ['open-code', 'opencode'],
   ['omp', 'omp'],
   ['oh-my-pi', 'omp'],
 ]);
 
-export const TOOLS_FLAG_HELP = 'all|none|claude,codex,opencode,omp';
+export const TOOLS_FLAG_HELP = 'all|none|claude,codex,omp';
 
 export function parseToolsArg(argv = []) {
   let toolsArg = null;
@@ -137,6 +125,15 @@ export function resolveOptionalToolKeys(toolsArg) {
       );
     }
 
+    // legacy: 'opencode'/'open-code' were adapters until the OpenCode lane was removed.
+    // Warn + ignore so `--tools=opencode` still exits 0 instead of hard-failing.
+    if (normalizedToken === 'opencode' || normalizedToken === 'open-code') {
+      console.warn(
+        `[UKit] '${token}' was removed as a UKit adapter — ignoring. Allowed: ${TOOLS_FLAG_HELP}`,
+      );
+      continue;
+    }
+
     const normalizedTool = TOOL_ALIASES.get(normalizedToken) ?? null;
     if (!normalizedTool) {
       throw new Error(`Unknown tool '${token}'. Allowed: ${TOOLS_FLAG_HELP}`);

package/src/cli/commands/doctor.js

@@ -24,6 +24,7 @@ import {
 } from '../../core/projectImportant.js';
 import { runDocContractChecks } from '../../core/docContracts.js';
 import { inspectUnattendedMode } from '../../core/unattendedDoctor.js';
+import { findOpencodeArtifacts, opencodeSteerMessage } from '../../core/opencodeSteer.js';
 
 export const DOCTOR_HELP_FLAGS = new Set(['--help', '-h']);
 const KNOWN_FLAGS = new Set([...DOCTOR_HELP_FLAGS, '--skills', '--gateway', '--docs']);
@@ -203,13 +204,16 @@ export async function runDoctor({ packageRoot, projectRoot, argv = [], homeDir =
       remedy: INSTALL_REPAIR,
     });
   }
-  if (piWiring.opencode.installed) {
+  // SPEC FR-007 — a leftover opencode.json is a WARN row: advisory only, never a
+  // FAIL, never auto-fixed. Removal is the user's manual action.
+  const opencodeArtifacts = findOpencodeArtifacts(projectRoot);
+  if (opencodeArtifacts.length > 0) {
     projectChecks.push({
-      label: 'OpenCode fallback present (AGENTS.md owner-instructions section)',
-      passed: piWiring.opencode.wired,
-      failed: !piWiring.opencode.wired,
-      remediationClass: 'install-repairable',
-      remedy: INSTALL_REPAIR,
+      label: opencodeSteerMessage(opencodeArtifacts),
+      passed: true,
+      severity: 'warning',
+      remediationClass: 'advisory',
+      remedy: 'Remove it manually: rm opencode.json',
     });
   }
 
@@ -252,7 +256,11 @@ export async function runDoctor({ packageRoot, projectRoot, argv = [], homeDir =
   console.log('[UKit] Project rules checks:');
   for (const check of projectChecks) {
     if (check.applicable === false) continue;
-    console.log(`[UKit]   ${ok(check.passed)}  ${check.label}`);
+    const severity = check.severity === 'warning' ? ' (warning)' : '';
+    console.log(`[UKit]   ${ok(check.passed)}  ${check.label}${severity}`);
+    if (check.remedy && check.severity === 'warning') {
+      console.log(`[UKit]      remedy: ${check.remedy}`);
+    }
   }
 
   // TASK-011 / SPEC §8 (FR-007): unattended-mode checks — same remediation-class

package/src/cli/index.js

@@ -117,8 +117,8 @@ export async function runCli({ argv, packageRoot, projectRoot, packageVersion })
     console.log('');
     console.log('Options (install / diff):');
     console.log('  --tools=<list>     Comma-separated adapters to enable (default: all)');
-    console.log('                     Values: claude, codex, opencode, omp');
-    console.log('                     Example: ukit install --tools=claude,codex,opencode,omp');
+    console.log('                     Values: claude, codex, omp');
+    console.log('                     Example: ukit install --tools=claude,codex,omp');
     console.log('                     Use --tools=none to disable all optional adapters');
     console.log('  --with-codegraph   Enable CodeGraph MCP integration in CLAUDE.md');
     console.log('                     Adds routing rules that prefer codegraph_* tools for');

package/src/core/opencodeSteer.js

@@ -0,0 +1,37 @@
+// SPEC FR-004 — OpenCode support was removed. Any leftover artifact is a user
+// file: UKit never installs, repairs, merges, or deletes it (only `ukit
+// uninstall` may). This helper is the single source of truth for detection and
+// the canonical steer message shared by projectImportant, status, and doctor.
+
+import { existsSync } from 'node:fs';
+import path from 'node:path';
+
+// Extensible list — detection is one existsSync per known filename, no parsing.
+const OPENCODE_ARTIFACT_NAMES = ['opencode.json'];
+
+/**
+ * @param {string} projectRoot
+ * @returns {string[]} absolute paths of detected OpenCode artifacts
+ */
+export function findOpencodeArtifacts(projectRoot) {
+  return OPENCODE_ARTIFACT_NAMES
+    .map((name) => path.join(projectRoot, name))
+    .filter((candidate) => existsSync(candidate));
+}
+
+/**
+ * Canonical steer copy. Callers must print this verbatim so every surface
+ * (status line, doctor WARN row, stderr warning) reads identically.
+ *
+ * @param {string[]} paths detected artifact paths from findOpencodeArtifacts
+ * @returns {string}
+ */
+export function opencodeSteerMessage(paths = []) {
+  const names = (Array.isArray(paths) ? paths : [])
+    .map((entry) => path.basename(String(entry)))
+    .filter(Boolean);
+  const fileList = names.length > 0 ? names.join(' ') : 'opencode.json';
+  return `[ukit] ${fileList} detected — OpenCode is no longer supported. `
+    + `ukit will not install, repair, or merge this file. `
+    + `Remove it manually: rm ${fileList}`;
+}

package/src/core/projectImportant.js

@@ -25,6 +25,7 @@ import {
   isSensitiveDataGateEnabled,
   loadSensitiveAllowlist,
 } from './sensitiveValueScanner.js';
+import { findOpencodeArtifacts, opencodeSteerMessage } from './opencodeSteer.js';
 
 export const PROJECT_IMPORTANT_FILENAME = 'PROJECT_IMPORTANT.md';
 export const PROJECT_IMPORTANT_CODEPOINT_LIMIT = 6000;
@@ -338,12 +339,14 @@ export async function renderProjectImportantForProject(projectRoot, options = {}
  * TASK-044 — adapter wiring facts for status/doctor (spec §13/§14).
  *
  * Returns install + wiring facts only — never file bodies. Claude wiring is
- * always expected; omp/Codex/OpenCode are checked only when installed, and an
- * absent adapter is `installed:false`, never a failure.
+ * always expected; omp/Codex are checked only when installed, and an
+ * absent adapter is `installed:false`, never a failure. A leftover
+ * `opencode.json` is no longer an adapter signal — it is detected and
+ * reported via `opencodeArtifacts` + a steer warning, never wired.
  *
  * @param {string} projectRoot
  * @param {{ trackedPaths?: string[] }} [options] install.json file list, if known
- * @returns {Promise<{claude: object, omp: object, codex: object, opencode: object, agentsFallback: boolean}>}
+ * @returns {Promise<{claude: object, omp: object, codex: object, opencodeArtifacts: string[], agentsFallback: boolean}>}
  */
 export async function inspectProjectImportantWiring(projectRoot, options = {}) {
   const trackedPaths = Array.isArray(options.trackedPaths) ? options.trackedPaths : [];
@@ -397,7 +400,7 @@ export async function inspectProjectImportantWiring(projectRoot, options = {}) {
     wired: runtimeModule && hookInstalled && hookExecutable && settingsWired,
   };
 
-  // --- AGENTS.md owner-instructions pointer (shared Codex/OpenCode fallback) ---
+  // --- AGENTS.md owner-instructions pointer (shared Codex fallback) ---
   let agentsFallback = false;
   try {
     const agents = await readFile(agentsPath, 'utf8');
@@ -425,9 +428,12 @@ export async function inspectProjectImportantWiring(projectRoot, options = {}) {
   const codexInstalled = tracked('.codex/') || await exists(path.join(projectRoot, '.codex', 'settings.json'));
   const codex = { installed: codexInstalled, wired: codexInstalled && agentsFallback };
 
-  // --- OpenCode fallback (only when opencode.json tracked/installed) ---
-  const opencodeInstalled = tracked('opencode.json') || await exists(path.join(projectRoot, 'opencode.json'));
-  const opencode = { installed: opencodeInstalled, wired: opencodeInstalled && agentsFallback };
+  // --- OpenCode steer (SPEC FR-005) — a leftover opencode.json is a user file:
+  // never wired, never installed/repaired. Detect it and steer toward removal.
+  const opencodeArtifacts = findOpencodeArtifacts(projectRoot);
+  if (opencodeArtifacts.length > 0) {
+    console.warn(opencodeSteerMessage(opencodeArtifacts));
+  }
 
-  return { claude, omp, codex, opencode, agentsFallback };
+  return { claude, omp, codex, opencodeArtifacts, agentsFallback };
 }

package/src/core/runInstallPipeline.js

@@ -15,6 +15,7 @@ import { ensureGitignore } from './ensureGitignore.js';
 import { repairBrokenHooks } from './repairBrokenHooks.js';
 import { applyGatewayResilienceEnv } from './gatewayResilienceEnv.js';
 import { cleanupEmptyParents, readJsonIfExists, removeFileOrLinkOnly, resolveProjectRelativePath } from './fileOps.js';
+import { findOpencodeArtifacts, opencodeSteerMessage } from './opencodeSteer.js';
 
 const AUTO_PRUNE_OBSOLETE_PREFIXES = [
   '.claude/skills/',
@@ -261,6 +262,15 @@ export async function runInstallPipeline({
   const installMetaData = await readJsonIfExists(pathConfig.installMetaPath);
   const isReinstall = Boolean(installMetaData?.tool === 'ukit');
 
+  // SPEC FR-004/§security — a leftover opencode.json is a user file: detect and
+  // steer toward manual removal. Install never writes, merges, repairs, or
+  // deletes it (obsolete-path pruning excludes it; only `ukit uninstall` may
+  // remove it).
+  const opencodeArtifacts = findOpencodeArtifacts(pathConfig.projectRoot);
+  if (opencodeArtifacts.length > 0) {
+    console.warn(opencodeSteerMessage(opencodeArtifacts));
+  }
+
   if (isReinstall && !dryRun) {
     await migrateLegacyRuntimeRoot(pathConfig.projectRoot);
   }

package/src/core/runtimeConfig.js

@@ -9,8 +9,10 @@ const require = createRequire(import.meta.url);
 const { version: PACKAGE_VERSION } = require('../../package.json');
 
 // legacy: 'antigravity' stays valid — Antigravity adapter removed in v2.2.0, but an existing
-// config.json naming it must not become invalid.
-const VALID_AGENTS = new Set(['claude-code', 'codex', 'antigravity', 'opencode', 'omp']);
+// config.json naming it must not become invalid. 'opencode' is NOT kept valid here: the
+// OpenCode lane was removed this cycle, so stale 'opencode' values are normalized away
+// (strip + console.warn) in normalizeLegacyOpencodeEntries instead of hard-failing.
+const VALID_AGENTS = new Set(['claude-code', 'codex', 'antigravity', 'omp']);
 
 // Reserved permission modes for orchestration.permissionMode (single source of truth).
 // 'unattended' is the only mode implemented this cycle; interactive/safe-auto are reserved.
@@ -90,7 +92,6 @@ export function buildDefaultRuntimeConfig(overrides = {}) {
         mustNotBlockMainTask: true,
         targets: {
           claude: { autoCompact: true, mode: 'precompact-reinject', preserveExistingHooks: true },
-          opencode: { autoCompact: true, mode: 'native-auto-prune', preserveExistingCompaction: true },
           // 'precompact-reinject' like claude, NOT codex's 'soft-handoff': omp exposes real
           // pre-compaction hook events, so there is nothing to work around with a state file.
           omp: { autoCompact: true, mode: 'precompact-reinject', preserveExistingHooks: true },
@@ -653,6 +654,37 @@ export function validateRuntimeConfig(config) {
   };
 }
 
+// OpenCode was removed as an adapter this cycle. Stale 'opencode' values inside
+// validated agent lists/keys are stripped with a steer warning — never a hard-fail —
+// so an old config.json still loads instead of falling back to defaults.
+function normalizeLegacyOpencodeEntries(config) {
+  const warned = { value: false };
+  const warnOnce = () => {
+    if (!warned.value) {
+      console.warn(`[UKit] Removed stale 'opencode' agent entries from UKit runtime config — OpenCode is no longer supported.`);
+      warned.value = true;
+    }
+  };
+
+  if (isPlainObject(config) && config.agent === 'opencode') {
+    warnOnce();
+    config.agent = 'claude-code';
+  }
+
+  if (isPlainObject(config) && Array.isArray(config.agents) && config.agents.includes('opencode')) {
+    warnOnce();
+    config.agents = config.agents.filter((agent) => agent !== 'opencode');
+  }
+
+  const targets = config?.compact?.agentContext?.targets;
+  if (isPlainObject(targets) && 'opencode' in targets) {
+    warnOnce();
+    delete targets.opencode;
+  }
+
+  return config;
+}
+
 export async function inspectRuntimeConfig(projectRoot) {
   const runtimePaths = buildRuntimePaths(projectRoot);
   let exists = false;
@@ -670,7 +702,7 @@ export async function inspectRuntimeConfig(projectRoot) {
     }
   }
 
-  const config = buildDefaultRuntimeConfig(rawConfig ?? {});
+  const config = normalizeLegacyOpencodeEntries(buildDefaultRuntimeConfig(rawConfig ?? {}));
   const validation = validateRuntimeConfig(config);
   const errors = [
     ...(parseError ? [`config.json parse error: ${parseError}`] : []),

package/src/core/status.js

@@ -8,13 +8,13 @@ import { countMemoryItems } from './memory/store.js';
 import { buildCompactPressureState } from './compact/threshold.js';
 import { detectProjectContext } from '../context/detectProjectContext.js';
 import { inspectProjectImportant, inspectProjectImportantWiring } from './projectImportant.js';
+import { opencodeSteerMessage } from './opencodeSteer.js';
 
 function formatPrimaryAgent(agentKey) {
   if (agentKey === 'claude-code') return 'Claude Code';
   if (agentKey === 'codex') return 'OpenAI Codex';
   // legacy: Antigravity adapter removed in this release; existing config.json may still name it
   if (agentKey === 'antigravity') return 'Antigravity';
-  if (agentKey === 'opencode') return 'OpenCode';
   if (agentKey === 'omp') return 'Oh My Pi (omp)';
   return agentKey || 'Unknown';
 }
@@ -24,7 +24,6 @@ async function detectAdapterLabels(projectRoot, primaryAgent) {
 
   const checks = [
     { label: 'OpenAI Codex', path: '.codex/settings.json' },
-    { label: 'OpenCode', path: 'opencode.json' },
     { label: 'Oh My Pi (omp)', path: '.omp/config.yml' },
   ];
 
@@ -171,8 +170,11 @@ export async function buildStatusReport(projectRoot) {
       claude: piWiring.claude.wired,
       omp: piWiring.omp.installed ? piWiring.omp.wired : undefined,
       codex: piWiring.codex.installed ? piWiring.codex.wired : undefined,
-      opencode: piWiring.opencode.installed ? piWiring.opencode.wired : undefined,
     },
+    // SPEC FR-006 — leftover opencode.json is a user file: steer, never wire.
+    opencodeSteer: piWiring.opencodeArtifacts.length > 0
+      ? opencodeSteerMessage(piWiring.opencodeArtifacts)
+      : null,
   };
 
   return {
@@ -226,7 +228,6 @@ function formatProjectImportantWiring(pi) {
   const parts = [`Claude runtime ${mark(pi.wiring.claude)}`];
   if (pi.wiring.omp !== undefined) parts.push(`omp runtime ${mark(pi.wiring.omp)}`);
   if (pi.wiring.codex !== undefined) parts.push(`Codex fallback ${mark(pi.wiring.codex)}`);
-  if (pi.wiring.opencode !== undefined) parts.push(`OpenCode fallback ${mark(pi.wiring.opencode)}`);
   return parts.join(' / ');
 }
 
@@ -237,6 +238,7 @@ export function formatStatusReport(report) {
     `${padLabel('Project')} ${report.projectName}`,
     `${padLabel('Project rules')} ${formatProjectImportantRules(report.projectImportant)}`,
     `${padLabel('Rule wiring')} ${formatProjectImportantWiring(report.projectImportant)}`,
+    ...(report.projectImportant.opencodeSteer ? [report.projectImportant.opencodeSteer] : []),
     `${padLabel('Adapters')} ${report.adapters.join(', ')}`,
     `${padLabel('Memory items')} ${report.memoryCounts.projectCount} project / ${report.memoryCounts.sessionCount} session / ${report.memoryCounts.userCount} user`,
     `${padLabel('Last compact')} ${report.lastCompact}`,

package/src/manifest/selectItems.js

@@ -5,7 +5,6 @@ export const OPTIONAL_ADAPTER_ITEM_IDS = new Set([
   'multi-codex-settings-json',
   'multi-codex-settings-local',
   'multi-codex-agents-link',
-  'multi-opencode-config',
   'multi-omp-agents-md',
   'multi-omp-rules',
   'multi-omp-config',

package/templates/.claude/agents/bug-debugger.md

@@ -50,7 +50,7 @@ Systematic debugging — understand before fixing.
 
 ```
 STATUS: DONE | BLOCKED | PARTIAL
-EXECUTOR_TOOL: [claude-code | kilo-code | codex | opencode | other]
+EXECUTOR_TOOL: [claude-code | kilo-code | codex | other]
 EXECUTOR_MODEL: [exact model name you are running as. "unknown" if you cannot tell.]
 EXECUTOR_SUBAGENT: [subagent name within your host, if any, else "-"]
 SUMMARY: [1-2 sentences — root cause and fix]

package/templates/.claude/agents/feature-implementer.md

@@ -72,7 +72,7 @@ and even then, report it, don't ask about it.
 
 ```
 STATUS: DONE | BLOCKED | PARTIAL
-EXECUTOR_TOOL: [claude-code | kilo-code | codex | opencode | other]
+EXECUTOR_TOOL: [claude-code | kilo-code | codex | other]
 EXECUTOR_MODEL: [exact model name you are running as — e.g. unic-code, claude-sonnet-4-5, gpt-5-mini. If you truly cannot tell, write "unknown" — reviewer treats unknown as suspicious and asks the human to confirm.]
 EXECUTOR_SUBAGENT: [name of the subagent you are, if your host has multiple — e.g. "Kilo:code", "Claude:feature-implementer". Otherwise "-".]
 SUMMARY: [1-2 sentences of what was implemented]

package/templates/.claude/agents/ukit-small-task-maintainer.md

@@ -24,7 +24,7 @@ You are UKit's internal small-task maintainer. You run as a sidecar/parallel/non
 - Summarizing `docs/STATUS.md`, `docs/TASKS.md`, `docs/WORKLOG.md`, or memory snippets into compact handoff notes.
 - Deciding fast vs slow lane, safe vs risky lane, whether a skill should be activated, and whether the current prompt has enough steps planned.
 - Choosing compact-now vs compact-later and summarize-vs-keep-detail for UKit context hygiene.
-- Keeping agent context compact without removing existing lanes: Claude PreCompact/reinject stays active, OpenCode native auto/prune compaction stays active, and Codex Desktop soft handoffs use `compact.codexContext.compactTarget` (default 150 lines; preferred 120-150; hard max 170) while preserving critical state.
+- Keeping agent context compact without removing existing lanes: Claude PreCompact/reinject stays active and Codex Desktop soft handoffs use `compact.codexContext.compactTarget` (default 150 lines; preferred 120-150; hard max 170) while preserving critical state.
 - Small, reversible UKit runtime maintenance decisions.
 
 ## Never Use For
@@ -45,7 +45,7 @@ You are UKit's internal small-task maintainer. You run as a sidecar/parallel/non
 5. If safe, make the smallest cleanup/summary/classification/routing decision.
 6. Follow the CoDev priority order: output quality first, then safety, then speed, then token discipline. Spend tokens/steps when they protect correctness.
 7. Use small step budgets: trivial = 1 step, simple = up to 2 steps, non-trivial = up to 4 planned steps before the main model reassesses.
-8. For long sessions, preserve existing Claude/OpenCode auto-compact behavior; for Codex Desktop, target a compact handoff of about 120-150 lines, never above 170 lines for this lane, not a full transcript replay.
+8. For long sessions, preserve existing Claude auto-compact behavior; for Codex Desktop, target a compact handoff of about 120-150 lines, never above 170 lines for this lane, not a full transcript replay.
 9. Preserve user-owned content; do not delete unclear tasks.
 10. Report exactly what changed or what decision was made.
 

package/templates/.claude/commands/ukit/handoff-create.md

@@ -1,7 +1,7 @@
 # /ukit:handoff-create — Phase 1 + 2: Plan
 
 **Role: PLANNER**
-**Tool: any** (Claude Code / Codex / omp / Kilo / OpenCode — your choice)
+**Tool: any** (Claude Code / Codex / omp / Kilo — your choice)
 **Model split:**
 - Read/understand → lite model (haiku · unic-lite · cheapest available)
 - Write plan + tasks → strong model (Opus · unic-smart · strongest available)
@@ -49,7 +49,7 @@ After the batched call, run Steps 1–2.5 to completion without further question
 4. Read `docs/AI_HANDOFF/tasks/_TEMPLATE.md` → task file structure
 5. Return a compact summary. Do NOT write anything yet.
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the lite model, run the steps above yourself, keep the summary in context.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the lite model, run the steps above yourself, keep the summary in context.
 
 ---
 
@@ -122,7 +122,7 @@ The planner agent does the following (use Step 1 summary — do NOT re-read file
 
 8. Report: task IDs, dependency graph, any `needs_breakdown` + reason
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the strong model, execute steps 1–7 above yourself.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the strong model, execute steps 1–7 above yourself.
 
 ---
 
@@ -143,7 +143,7 @@ intact. What it no longer does is hand a stalled plan back and wait.
 2. `Issues Found` → route back to Step 2: planner revises `PLAN.md` and the affected `TASK-xxx.md` files to address every finding, then re-submit for another Step 2.5 review (this becomes the next round). Do NOT commit or hand off to executor on `Issues Found`.
 3. `Approved` → append `PLAN_REVIEW: Approved by <reviewer model>` to PLAN.md's `## Planner Report` footer, then proceed.
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the strong model in a **separate** chat/session from Step 2, paste PLAN.md, review using the Spec/Plan Review checklist in `.claude/agents/code-reviewer.md`.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the strong model in a **separate** chat/session from Step 2, paste PLAN.md, review using the Spec/Plan Review checklist in `.claude/agents/code-reviewer.md`.
 
 ---
 

package/templates/.claude/commands/ukit/handoff-fullstack.md

@@ -1,7 +1,7 @@
 # /ukit:handoff-fullstack — Full Pipeline: Plan → Implement → Review
 
 **Role: ORCHESTRATOR**
-**Tool: any** (Claude Code / Codex / omp / Kilo / OpenCode — your choice)
+**Tool: any** (Claude Code / Codex / omp / Kilo — your choice)
 
 ## Model Split
 
@@ -186,7 +186,7 @@ counter and starts a new cycle automatically.
 
 Return a compact summary. Do NOT write any files yet.
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the lite model and run P1 yourself.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the lite model and run P1 yourself.
 
 ### P2 — Write PLAN.md + task files (strong model)
 
@@ -270,7 +270,7 @@ is written. What it no longer does is hand a stalled plan back to a human who is
 2. `Issues Found` → route back to P2: `handoff-planner` revises `PLAN.md` and the affected `TASK-xxx.md` files to address every finding, then re-submit for one more P2.5 review round — subject to the loop cap above.
 3. `Approved` → append `PLAN_REVIEW: Approved by <reviewer model>` to PLAN.md's `## Planner Report` footer, then proceed to P3.
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the strong model in a **separate** chat/session from P2, paste PLAN.md, review using the Spec/Plan Review checklist in `.claude/agents/code-reviewer.md`.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the strong model in a **separate** chat/session from P2, paste PLAN.md, review using the Spec/Plan Review checklist in `.claude/agents/code-reviewer.md`.
 
 ### P3 — Commit the plan (lite model)
 
@@ -281,7 +281,7 @@ git add docs/AI_HANDOFF/ && git commit -m "handoff: plan — <goal>"
 
 Replace `<goal>` with the one-sentence goal from ACTIVE.md. This locks the plan in git before any implementation begins.
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the lite model for P2/P3.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the lite model for P2/P3.
 
 ---
 

package/templates/.claude/commands/ukit/handoff-implement.md

@@ -1,7 +1,7 @@
 # /ukit:handoff-implement — Phase 3: Execute
 
 **Role: EXECUTOR (orchestrated)**
-**Tool: any** (Claude Code / Codex / omp / Kilo / OpenCode — your choice)
+**Tool: any** (Claude Code / Codex / omp / Kilo — your choice)
 **Model: code model** (Sonnet · unic-code · cheap-smart)
 
 > **omp model tiers:** the tiers above map to `.omp/config.yml`'s `modelRoles`, referenced from agent frontmatter as `@smol` / `@default` / `@slow`.
@@ -170,7 +170,7 @@ disk in the task file, which is where the reviewer reads them from. Pasting them
 second time is what blows up the orchestrator's context window and kills the run mid-wave.
 ```
 
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** open each task in a separate session with the code model.
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** open each task in a separate session with the code model.
 
 ### 3c — Orchestrator: copy changes to main + IMMEDIATELY delete worktree
 

package/templates/.claude/commands/ukit/handoff-review.md

@@ -1,7 +1,7 @@
 # /ukit:handoff-review — Phase 4: Review
 
 **Role: REVIEWER**
-**Tool: any** (Claude Code / Codex / omp / Kilo / OpenCode — your choice)
+**Tool: any** (Claude Code / Codex / omp / Kilo — your choice)
 **Model: strong model, MUST differ from executor** (Opus · unic-smart · strongest available)
 
 > **omp model tiers:** the tiers above map to `.omp/config.yml`'s `modelRoles`, referenced from agent frontmatter as `@smol` / `@default` / `@slow`.
@@ -191,4 +191,4 @@ This is the only place this command may ask for a compaction. All state lives in
 `INDEX.md` and `RUN.md`, so a compacted or fresh session resumes with nothing lost.
 
 > Orchestrator (this session) handles Step 1, 2e, and 3 directly — those are not delegated.
-> **For the human operator, on a tool with no agent support (Codex, OpenCode) — not an instruction to the model:** manually switch to the strong model, run 2a–2d per task (one session at a time).
+> **For the human operator, on a tool with no agent support (Codex) — not an instruction to the model:** manually switch to the strong model, run 2a–2d per task (one session at a time).

package/templates/.claude/skills/update-status/SKILL.md

@@ -48,7 +48,7 @@ When updating meaningful state, update the header:
 ## Freshness
 
 - Last meaningful update: YYYY-MM-DD HH:mm
-- Updated by: Claude / Codex / OpenCode / human
+- Updated by: Claude / Codex / human
 - Status confidence: high / medium / low
 - Stale after: 72h
 ```

package/templates/.claude/ukit/index/route-task.mjs

@@ -816,7 +816,12 @@ async function main() {
   const lastPrompt = readFlagValue(args, '--last-prompt')
     ?? previousState?.routingContext?.lastExplicitUserPromptText
     ?? '';
-  const adapter = readFlagValue(args, '--adapter') ?? inferAdapterFromInvocation(process.argv[1] || '');
+  const KNOWN_ADAPTERS = new Set(['claude', 'codex', 'antigravity']);
+  const rawAdapter = readFlagValue(args, '--adapter') ?? inferAdapterFromInvocation(process.argv[1] || '');
+  const adapter = KNOWN_ADAPTERS.has(rawAdapter) ? rawAdapter : 'claude';
+  if (rawAdapter !== adapter) {
+    console.error(`route-task: unknown --adapter "${rawAdapter}" — falling back to claude`);
+  }
   const commandNamespace = adapter === 'codex'
     ? '.codex'
     : (adapter === 'antigravity' ? '.antigravity' : '.claude');
@@ -826,7 +831,7 @@ async function main() {
     .trim();
 
   if (!promptText && !commandText && !targetFile) {
-    console.error('Usage: node .codex/ukit/index/route-task.mjs "<prompt>" [--tool-command <cmd>] [--target <file>] [--type trivial|simple|non-trivial] [--session-id <id>] [--adapter codex|claude|antigravity|opencode]');
+    console.error('Usage: node .codex/ukit/index/route-task.mjs "<prompt>" [--tool-command <cmd>] [--target <file>] [--type trivial|simple|non-trivial] [--session-id <id>] [--adapter codex|claude|antigravity]');
     process.exitCode = 1;
     return;
   }

package/templates/.codex/README.md

@@ -8,7 +8,7 @@ Auto-generated by UKit for OpenAI Codex.
 ## Core UKit Rule
 
 - Human-facing workflow should optimize for one remembered command: `ukit install`.
-- After install, normal work should feel natural inside **Claude/Codex/OpenCode/omp**.
+- After install, normal work should feel natural inside **Claude/Codex/omp**.
 - **Quality first, then speed, then token discipline**: do not waste reads, logs, or repeated helper output.
 - **Never stop after read-only steps.** For implement/apply/fix requests, continue to actual Edit/Write and verification in the same turn.
 

package/templates/.codex/settings.json

@@ -233,7 +233,6 @@
       },
       "agentContext": {
         "preserveClaudePreCompact": true,
-        "preserveOpenCodeCompaction": true,
         "codexSoftHandoff": {
           "autoCompact": true,
           "budgetTokens": 100000,

package/templates/.omp/agents/bug-debugger.md

@@ -49,7 +49,7 @@ Systematic debugging — understand before fixing.
 
 ```
 STATUS: DONE | BLOCKED | PARTIAL
-EXECUTOR_TOOL: [claude-code | kilo-code | codex | opencode | other]
+EXECUTOR_TOOL: [claude-code | kilo-code | codex | other]
 EXECUTOR_MODEL: [exact model name you are running as. "unknown" if you cannot tell.]
 EXECUTOR_SUBAGENT: [subagent name within your host, if any, else "-"]
 SUMMARY: [1-2 sentences — root cause and fix]

package/templates/.omp/agents/feature-implementer.md

@@ -71,7 +71,7 @@ and even then, report it, don't ask about it.
 
 ```
 STATUS: DONE | BLOCKED | PARTIAL
-EXECUTOR_TOOL: [claude-code | kilo-code | codex | opencode | other]
+EXECUTOR_TOOL: [claude-code | kilo-code | codex | other]
 EXECUTOR_MODEL: [exact model name you are running as — e.g. unic-code, claude-sonnet-4-5, gpt-5-mini. If you truly cannot tell, write "unknown" — reviewer treats unknown as suspicious and asks the human to confirm.]
 EXECUTOR_SUBAGENT: [name of the subagent you are, if your host has multiple — e.g. "Kilo:code", "Claude:feature-implementer". Otherwise "-".]
 SUMMARY: [1-2 sentences of what was implemented]

package/templates/.omp/agents/ukit-small-task-maintainer.md

@@ -24,7 +24,7 @@ You are UKit's internal small-task maintainer. You run as a sidecar/parallel/non
 - Summarizing `docs/STATUS.md`, `docs/TASKS.md`, `docs/WORKLOG.md`, or memory snippets into compact handoff notes.
 - Deciding fast vs slow lane, safe vs risky lane, whether a skill should be activated, and whether the current prompt has enough steps planned.
 - Choosing compact-now vs compact-later and summarize-vs-keep-detail for UKit context hygiene.
-- Keeping agent context compact without removing existing lanes: Claude PreCompact/reinject stays active, OpenCode native auto/prune compaction stays active, and Codex Desktop soft handoffs use `compact.codexContext.compactTarget` (default 150 lines; preferred 120-150; hard max 170) while preserving critical state.
+- Keeping agent context compact without removing existing lanes: Claude PreCompact/reinject stays active and Codex Desktop soft handoffs use `compact.codexContext.compactTarget` (default 150 lines; preferred 120-150; hard max 170) while preserving critical state.
 - Small, reversible UKit runtime maintenance decisions.
 
 ## Never Use For
@@ -45,7 +45,7 @@ You are UKit's internal small-task maintainer. You run as a sidecar/parallel/non
 5. If safe, make the smallest cleanup/summary/classification/routing decision.
 6. Follow the CoDev priority order: output quality first, then safety, then speed, then token discipline. Spend tokens/steps when they protect correctness.
 7. Use small step budgets: trivial = 1 step, simple = up to 2 steps, non-trivial = up to 4 planned steps before the main model reassesses.
-8. For long sessions, preserve existing Claude/OpenCode auto-compact behavior; for Codex Desktop, target a compact handoff of about 120-150 lines, never above 170 lines for this lane, not a full transcript replay.
+8. For long sessions, preserve existing Claude auto-compact behavior; for Codex Desktop, target a compact handoff of about 120-150 lines, never above 170 lines for this lane, not a full transcript replay.
 9. Preserve user-owned content; do not delete unclear tasks.
 10. Report exactly what changed or what decision was made.
 

package/templates/AGENTS.md

@@ -4,7 +4,7 @@
 ## Core Rule
 <!-- RULES: CORE-01 CORE-02 -->
 - Human-facing UKit workflow collapses to one remembered command: `ukit install`.
-- After install, default to natural-language work inside **Claude Code / Codex / OpenCode / omp**.
+- After install, default to natural-language work inside **Claude Code / Codex / omp**.
 - **Quality first, then speed, then token discipline.**
 - **Never stop after read-only steps** — implement/apply/fix requests continue to Edit/Write + verification in the same turn.
 
@@ -140,25 +140,16 @@ CHỈ kích hoạt khi task đi qua `docs/AI_HANDOFF/` (user nói "execute task
 - Contract map: `tiny-fix` → lite · `local-fix`, `local-build`, `shared-edit`, `find-cause`, `map-impact` → code · `review-release` → smart. Escalation: same file/symbol failing `debugLoopThreshold` (default 2) times routes the next attempt one tier higher, capped at `smart`.
 - `unic-vision` is a capability lane, not a cost tier — unverified vision must never guess image contents; route images to `ukit-vision-analyst`. Harness table: `docs/UKIT_INTERNALS.md`.
 
-## Session Start — OpenCode
+## Project Owner Instructions — Codex
 
-<!-- RULE: HOST-OC-01 -->
-At the start of every OpenCode session, before working on the first task:
-1. Reuse matching route hints from `.claude/ukit/skill-router-state.json`; scan `.claude/skills/` (listing only) and read a SKILL.md only when a task triggers it.
-2. For every non-trivial task, run `/ukit-route <task summary>` for skill + context hints **before** writing code; if it names a skill, read that SKILL.md — do not skip. When `router.enabled: true` in `.ukit/storage/config.json`, prefer router output over guessing; treat it as internal continuation — continue to edit/verify, don't stop.
-
-## Project Owner Instructions — Codex and OpenCode
-
-When running in Codex or OpenCode, read and follow the root
+When running in Codex, read and follow the root
 `PROJECT_IMPORTANT.md` before doing project work. It is the canonical
 project-owner instruction source. Do not copy its contents into this file.
 If it is missing or unreadable, state that limitation and continue with the
 remaining project instructions.
-<!-- RULES: OWN-01 HOST-OWN-01 -->
+<!-- RULE: OWN-01 -->
 ## Skills
 - Canonical skills live in `.claude/skills/`; adapter mirrors may exist (`.codex/skills/` → symlink). **omp** reads `.claude/skills/` via its `claude` discovery provider.
-- **OpenCode**: reads `AGENTS.md` at session start only — it does NOT auto-load skills; the model must explicitly read the triggered SKILL.md.
-- `ukit-*` commands in `opencode.json` are internal helper entrypoints — never ask end users to run them.
 
 ## Project Snapshot
 - Project: {{project.name}} | Root: {{project.root}}