oh-my-customcodex 1.0.6 → 1.0.7

package/dist/cli/index.js

@@ -3091,7 +3091,7 @@ var init_package = __esm(() => {
     workspaces: [
       "packages/*"
     ],
-    version: "1.0.6",
+    version: "1.0.7",
     requiresCC: ">=2.1.121",
     claudeCode: {
       minimumVersion: "2.1.121",

package/dist/index.js

@@ -2316,7 +2316,7 @@ var package_default = {
   workspaces: [
     "packages/*"
   ],
-  version: "1.0.6",
+  version: "1.0.7",
   requiresCC: ">=2.1.121",
   claudeCode: {
     minimumVersion: "2.1.121",

package/package.json

@@ -3,7 +3,7 @@
   "workspaces": [
     "packages/*"
   ],
-  "version": "1.0.6",
+  "version": "1.0.7",
   "requiresCC": ">=2.1.121",
   "claudeCode": {
     "minimumVersion": "2.1.121",

package/templates/.claude/rules/MUST-agent-design.md

@@ -37,6 +37,9 @@ Fable 5 effort strategy: Fable 5 defaults to high effort; reserve `xhigh` for ca
 Claude Code v2.1.198+ makes the built-in Explore agent inherit the main session model (capped at Opus) instead of staying fixed to Haiku, and makes subagents plus compaction inherit the session's extended-thinking setting. This improves Claude-template delegation quality, but Codex-native subagents still follow the installed OMX role/model contract.
 
 Claude Code v2.1.199+ shows stderr for SessionStart/Setup/SubagentStart hooks that exit 2, improving hard-block/advisory hook debugging. It also loads up to five leading stacked slash-skill calls (for example `/skill-a /skill-b do X`), reducing context loss in chained compatibility skills such as `omcustomcodex:fsd`; keep using explicit `$skill`/namespaced surfaces in Codex sessions.
+
+Claude Code v2.1.201+ changes Claude Sonnet 5 harness-reminder delivery so rule reminders are no longer injected as mid-conversation `system` role messages. This is a Claude-template prompt-shape change only; PostCompact rule reinjection and Codex/OMX session continuity remain governed by this repository's normal hooks and state.
+
 Claude compatibility settings can declare up to three `fallbackModel` entries tried in order when the primary Claude model is overloaded or unavailable. `--fallback-model` also applies to interactive Claude sessions. Treat this as platform availability failover, not Codex-native model routing or outcome-based escalation. Claude Code v2.1.166+ also supports disabling default thinking with `MAX_THINKING_TOKENS=0`, `--thinking disabled`, or the per-model thinking toggle. Claude Code v2.1.169+ adds `--safe-mode` / `CLAUDE_CODE_SAFE_MODE` to disable customizations for regression isolation, plus `disableBundledSkills` / `CLAUDE_CODE_DISABLE_BUNDLED_SKILLS` to hide bundled skills/workflows/slash commands when they conflict with project skills. Codex-native agents continue to use the OMX model contract and `reasoning_effort` routing. Claude Code v2.1.172+ applies `availableModels` restrictions to subagent `model:` overrides, the agent dispatch model picker, and the advisor model; compatibility allowlists should account for version-specific IDs and 1M suffix handling. Claude Code v2.1.173+ auto-normalizes Fable 5 IDs with redundant `[1m]` suffixes. Claude Code v2.1.175+ adds `enforceAvailableModels`, which constrains the resolved Default model as well as subagent overrides and prevents user/project settings from widening a managed allowlist.
 -->
 
@@ -203,6 +206,8 @@ When spawning agents via the Agent tool, CC applies a default `mode` of `acceptE
 1. **Agent frontmatter `permissionMode`**: Declares the agent's intended permission level. CC respects this when the agent is spawned via Agent tool.
 2. **Agent tool `mode` parameter**: Overrides frontmatter at spawn time. Routing skills should pass this explicitly.
 3. **Recommendation**: For agents that modify files, set `permissionMode: bypassPermissions` in frontmatter if the project uses `bypassPermissions` mode.
+
+Claude Code v2.1.200+ displays `default` permission mode as `Manual` in CLI help and IDE surfaces, and accepts `--permission-mode manual` / `"defaultMode": "manual"` alongside `default` with the same behavior. This is a Claude compatibility label change only; Codex/OMX approval policy and sandbox state remain controlled by the current Codex runtime.
 -->
 
 <!-- DETAIL: Isolation/Token/Limitations/Escalation details

package/templates/.claude/rules/MUST-completion-verification.md

@@ -174,6 +174,8 @@ Origin: upstream #1443; prevents incomplete R017/deep-verify handoffs from being
 
 > **Claude Code v2.1.199+ compatibility**: subagent API errors are reported to the parent instead of being self-reported as success. This lowers false-success frequency, but R020 still requires canonical ground-truth verification (`git status`, grep, tests, validation scripts, registry/API checks) before accepting a subagent report.
 
+> **Claude Code v2.1.200+ compatibility**: subagents that hit a rate limit before emitting any output now return a clean failure instead of an empty result. This removes one silent-empty failure mode, but the R020 rule is unchanged: do not accept delegated completion without deterministic repository, test, validation, registry, or API evidence.
+
 ## Common False Completion Patterns — 7 anti-patterns including "Command executed" without exit code check, "Waiting for manual publish" when CI auto-publishes. See full table via Read tool.
 
 <!-- DETAIL: Common False Completion Patterns

package/templates/.claude/rules/MUST-orchestrator-coordination.md

@@ -258,6 +258,8 @@ Claude Code v2.1.174+ fixes background sessions inheriting another session's `AN
 
 Claude Code v2.1.198+ background agents can auto-commit, push, and open draft PRs after completing work in a worktree, and retry transient network aborts. Claude Code v2.1.199+ returns partial subagent work on rate-limit/server errors and fixes several background-daemon lifecycle races. These are Claude-template reliability improvements only; Codex/OMX workflows must still make write authority explicit, keep release branches auditable, and verify the repository state before claiming completion.
 
+Claude Code v2.1.200+ further hardens background sessions and agents: sleep/wake or stalled-session resumes no longer silently stop mid-turn, cancelled respawned turns are not replayed after Esc, stale `daemon.lock` files from crashed workers are handled, older reinstalled builds cannot hijack newer daemons, roster corruption no longer disables orphan cleanup permanently, older binaries preserve newer record fields, and daemon restarts preserve socket auth tokens. Treat this as Claude-template reliability only; Codex/OMX still requires explicit branch, PR, and working-tree evidence.
+
 ## Agent Capability Pre-Check
 
 Before delegating work, compare the task requirements with the target agent frontmatter:

package/templates/.claude/rules/MUST-permissions.md

@@ -55,6 +55,12 @@ Claude Code compatibility templates may rely on v2.1.166+ deny-rule glob behavio
 
 Use `"*"` deny rules only for Claude compatibility settings that intentionally enforce deny-by-default, then add explicit allow rules. This is separate from Codex/OMX sandbox policy and the advisory tool-tier table above.
 
+## Claude Permission Mode Manual Label and AskUserQuestion (v2.1.200+)
+
+Claude Code v2.1.200+ displays the existing `default` permission mode as `Manual` across CLI help and IDE surfaces. `--permission-mode manual` and `"defaultMode": "manual"` are accepted alongside `default` with identical behavior. Document both names in Claude compatibility guidance, but do not translate the UI label into Codex sandbox policy.
+
+The same Claude release changed `AskUserQuestion` so dialogs no longer auto-continue by default; idle-timeout continuation is now explicit `/config` opt-in. Autonomous flows such as `omcustomcodex:fsd` should avoid question tools unless truly blocked and should proceed on best judgment for low-risk reversible work. Non-array `disabledMcpServers` / `enabledMcpServers` values in `.claude.json` no longer crash Claude startup.
+
 ## Permission Request Format
 
 ```

package/templates/guides/claude-code/15-version-compatibility.md

@@ -2,6 +2,30 @@
 
 This guide records Claude Code release-note impact that affects the Claude compatibility template. The Codex-native runtime still uses `.codex/**` and OMX as the primary surface.
 
+## v2.1.201
+
+Reviewed: 2026-07-06.
+
+Source: upstream oh-my-customcode v1.1.7 / #1567.
+
+| Change | Impact on oh-my-customcodex | Action |
+|--------|------------------------------|--------|
+| Claude Sonnet 5 sessions no longer receive harness reminders as mid-conversation `system` role messages | Prompt-shape compatibility change for packaged Claude templates only. PostCompact rule reinjection and Codex/OMX continuity are unchanged. | Document in R006 and keep Codex-native routing/state on AGENTS.md plus OMX surfaces. |
+
+## v2.1.200
+
+Reviewed: 2026-07-06.
+
+Source: upstream oh-my-customcode v1.1.7 / #1567.
+
+| Change | Impact on oh-my-customcodex | Action |
+|--------|------------------------------|--------|
+| `default` permission mode is displayed as `Manual`; `manual` is accepted alongside `default` | Claude compatibility templates and operator docs should recognize both names as the same behavior. | Document in R002/R006; do not map the UI label onto Codex sandbox policy. |
+| `AskUserQuestion` no longer auto-continues by default | Unattended Claude compatibility flows can now block indefinitely if they ask questions without a real blocker. | Document in R002; FSD/auto-dev should continue using best judgment for low-risk reversible work. |
+| Subagents rate-limited before first output now return clean failure instead of an empty result | Reduces silent-empty delegated results, but does not remove the need for ground-truth verification. | Document in R020 and keep deterministic `git status`/validation/test/API checks. |
+| Background-session and daemon lifecycle fixes cover sleep/wake stalls, stale locks, daemon freshness, roster/orphan cleanup, record preservation, and socket auth token handling | Fewer Claude background-agent false blockers. | Document in R010; Codex/OMX release flow still verifies branch, PR, tag, registry, and working-tree state directly. |
+| `.claude.json` non-array MCP server lists no longer crash startup | Claude compatibility config is more forgiving. | Mention in R002 as Tier-6/MCP startup resilience; no Codex config schema change. |
+
 ## v2.1.199
 
 Reviewed: 2026-07-05.

package/templates/manifest.json

@@ -1,11 +1,11 @@
 {
-  "version": "1.0.6",
+  "version": "1.0.7",
   "requiresCC": ">=2.1.121",
   "claudeCode": {
     "minimumVersion": "2.1.121",
     "protectedPathBypassVersion": "2.1.126"
   },
-  "lastUpdated": "2026-07-05",
+  "lastUpdated": "2026-07-06",
   "components": [
     {
       "name": "rules",