release: v1.1.0 — cbepx fork of codex-plugin-cc

.claude-plugin/marketplace.json

@@ -1,17 +1,18 @@
 {
-  "name": "openai-codex",
+  "name": "cbepx",
   "owner": {
-    "name": "OpenAI"
+    "name": "CBEPX",
+    "url": "https://github.com/CBEPX"
   },
   "metadata": {
-    "description": "Codex plugins to use in Claude Code for delegation and code review.",
-    "version": "1.0.6"
+    "description": "CBEPX fork of the OpenAI Codex plugin for Claude Code: max/ultra effort, per-thread config overrides, gpt-5.6 aliases, rescue agent fixes.",
+    "version": "1.1.0"
   },
   "plugins": [
     {
       "name": "codex",
       "description": "Use Codex from Claude Code to review code or delegate tasks.",
-      "version": "1.0.6",
+      "version": "1.1.0",
       "author": {
         "name": "OpenAI"
       },

.github/workflows/pull-request-ci.yml

@@ -2,6 +2,8 @@ name: Pull Request CI
 
 on:
   pull_request:
+  push:
+    branches: [main, "release/**"]
 
 permissions:
   contents: read

CHANGELOG.md

@@ -0,0 +1,34 @@
+# Changelog
+
+## 1.1.0 — 2026-08-27
+
+Fork of [openai/codex-plugin-cc](https://github.com/openai/codex-plugin-cc) 1.0.6 (`db52e28`). Marketplace `cbepx`, plugin name unchanged (`codex`).
+
+### Merged from upstream pull requests
+- #616 accept `max` and `ultra` reasoning efforts
+- #688 resolve model aliases on `review` / `adversarial-review`
+- #426 `on-request` approval policy for `--write` task runs
+- #501 answer MCP elicitation requests instead of rejecting them
+- #608 rescue agent awaits the delegated result instead of returning a placeholder
+- #690 explicit Bash blocks in `status`/`result`/`cancel`/`transfer` commands (pass permission classifiers)
+- #547 unknown flags are CLI errors, never part of the prompt; `--help` prints usage and exits 0
+- #645 / #644 job records store resolved model/effort/sandbox; reasoning start is logged
+- #672 SessionStart hook timeout raised; #668 idempotent `CLAUDE_ENV_FILE` exports; #682 stop gate fails closed on malformed input; #396 `CODEX_REVIEW_GATE_MAX_ROUNDS`
+
+### Fork changes
+- Slash-command arguments reach the companion through a quoted heredoc on stdin (`--args-stdin`) instead of a shell string: Claude Code substitutes `$ARGUMENTS` before bash runs, so `$(...)`/backticks in a prompt used to execute on the host shell, outside Codex's sandbox. Rescue job ids are validated before use. `/codex:rescue` keeps the two channels separate — the request prose goes to `--prompt-file` through its own quoted heredoc so quotes, backslashes and newlines survive byte-exact, while `--args-stdin` carries only runtime flags — and randomizes both heredoc delimiters per call; the other seven command bodies keep the fixed `CODEX_ARGS` delimiter because their payload is only flags and job ids.
+- Approval requests (`execCommandApproval`, `applyPatchApproval`, `item/commandExecution/requestApproval`, `item/fileChange/requestApproval`, `item/permissions/requestApproval`) are answered with each type's refusal variant instead of a `-32601` protocol error that made `--write` turns fail or hang.
+- Background task records are written before the worker is spawned (a fast worker used to find no record and exit while the launch reported `queued`), and the worker reads the full request — including `--config` values — from a private one-shot `jobs/<id>.request.json` (mode 0600); the job record `status`/`result` echo keeps secret-looking config values redacted.
+- Model and reasoning effort are sent per thread via `thread/start.config` (`model`, `review_model` for native review, `model_reasoning_effort`); generic `--config` pairs are applied first, dedicated flags override them; `--effort` now works on `review` and `adversarial-review`.
+- Repeatable `--config key=value` on `task`, `review`, `adversarial-review` forwards any `config.toml` override to the thread (values are JSON-parsed; quote a literal string as `'"true"'`). Prompt-taking commands stop option parsing at the first positional, so prompt text like `ls -R` is never mis-parsed.
+- `--resume-last` opens a fresh app-server session (cold resume) so `--config`, sandbox and approval policy take effect, and never sends `model` on `thread/resume` (it would drop the persisted model); the resumed turn's model/effort ride on `turn/start`.
+- All MCP elicitations are declined (no operator is present); URL/form flows must be completed in an interactive Codex session.
+- `/codex:rescue` is synchronous by default without the `Agent` tool: `task --background` → `status --wait` in ≤9-minute slices (job id carried literally between Bash calls) → `result`; launch failures stop immediately with visible stderr; `Agent` only for `--background`; `--write` is never added unless the user explicitly asked Codex to modify files.
+- `/codex:rescue` asks before continuing an existing Codex thread (`Continue current Codex thread` / `Start a new Codex thread`) instead of resuming silently; its `allowed-tools` is now `Bash, AskUserQuestion, Agent` because the body is multi-command shell.
+- A resume refuses to start a second turn on a thread that a queued or running job is still using, including a job from another Claude session.
+- Model aliases: `sol`, `luna`, `terra`, `mini` (plus `spark`); rescue agent has no pinned `model:`; runtime skill mentions `$agent-compat:skill-router` for uncommon domains.
+- Stop-gate script timeout (13 min) is below the hook timeout (15 min); `spawnSync` uses `SIGKILL` and a 16 MiB buffer.
+- Hermetic test environment (`tests/test-env.mjs`); CI on push; `npm run build` type-checks the JSDoc.
+
+## 1.0.6 and earlier
+See upstream releases: https://github.com/openai/codex-plugin-cc/releases

README.md

@@ -1,5 +1,7 @@
 # Codex plugin for Claude Code
 
+> **CBEPX fork.** Install with `claude plugin marketplace add CBEPX/codex-plugin-cc` then `claude plugin install codex@cbepx`. Differences from upstream are listed in [CHANGELOG.md](CHANGELOG.md). Upstream: openai/codex-plugin-cc.
+
 Use Codex from inside Claude Code for code reviews or to delegate tasks to Codex.
 
 This plugin is for Claude Code users who want an easy way to start using Codex from the workflow
@@ -24,13 +26,13 @@ they already have.
 Add the marketplace in Claude Code:
 
 ```bash
-/plugin marketplace add openai/codex-plugin-cc
+/plugin marketplace add CBEPX/codex-plugin-cc
 ```
 
 Install the plugin:
 
 ```bash
-/plugin install codex@openai-codex
+/plugin install codex@cbepx
 ```
 
 Reload plugins:
@@ -159,7 +161,9 @@ Ask Codex to redesign the database connection to be more resilient.
 **Notes:**
 
 - if you do not pass `--model` or `--effort`, Codex chooses its own defaults.
-- if you say `spark`, the plugin maps that to `gpt-5.3-codex-spark`
+- `--effort` accepts `none`, `minimal`, `low`, `medium`, `high`, `xhigh`, `max`, and `ultra`. Which of those a given model actually supports is decided by Codex, not by the plugin — run `codex debug models` to see the reasoning levels each model advertises.
+- model aliases: `spark` -> `gpt-5.3-codex-spark`, `sol` -> `gpt-5.6-sol`, `luna` -> `gpt-5.6-luna`, `terra` -> `gpt-5.6-terra`, `mini` -> `gpt-5.4-mini`
+- `--config key=value` (repeatable, also on `/codex:review` and `/codex:adversarial-review`) forwards a `config.toml` override to the Codex thread, e.g. `--config model_provider=ollama`. On `--resume-last` the plugin opens a fresh app-server session (cold resume) so `--config` overrides, sandbox and approval policy take effect; model and effort for the resumed turn are sent on the turn, never on the resume request.
 - follow-up rescue requests can continue the latest Codex task in the repo
 
 ### `/codex:transfer`
@@ -236,6 +240,17 @@ When the review gate is enabled, the plugin uses a `Stop` hook to run a targeted
 > [!WARNING]
 > The review gate can create a long-running Claude/Codex loop and may drain usage limits quickly. Only enable it when you plan to actively monitor the session.
 
+#### Bounding the review gate
+
+By default the gate keeps blocking the stop until Codex is satisfied, which is what can create the loop above. Set `CODEX_REVIEW_GATE_MAX_ROUNDS` to cap how many consecutive gate rounds run in a single session before the stop is allowed through:
+
+```bash
+# allow at most 5 stop-gate review rounds per session, then let the stop proceed
+export CODEX_REVIEW_GATE_MAX_ROUNDS=5
+```
+
+When unset or `0`, the gate is unbounded (the previous behavior). The count is per session, increments on each blocked round (tracked via `stop_hook_active`), and resets once a stop is allowed or a fresh user turn begins.
+
 ## Typical Flows
 
 ### Review Before Shipping