
Ce Work Beta
Delegate implementation units to Codex CLI while Claude Code keeps planning, review, git, and orchestration when compound-engineering delegation is on.
Overview
ce-work-beta is an agent skill most often used in Build (also Ship review, Ship testing) that delegates implementation units to Codex CLI while the orchestrator agent keeps planning and git control.
Install
npx skills add https://github.com/everyinc/compound-engineering-plugin --skill ce-work-betaWhat is this skill?
- Routes implementation to codex exec when delegation_active is true
- Orchestrator retains planning, review, git operations, and batch orchestration
- work_delegate_decision ask mode presents Codex vs Claude Code with explicit recommendations
- auto mode states batch plan in one line: N units in X batch(es)
- User can override to Claude Code for trivial units or force Codex when recommended otherwise
Adoption & trust: 1.5k installs on skills.sh; 20.5k GitHub stars; 1/3 security scanners passed (skills.sh audits).
What problem does it solve?
You want Claude Code to plan and review but need heavier implementation passes farmed out without losing orchestration or batch discipline.
Who is it for?
Solo builders already on Claude Code plus Codex with compound-engineering work flows and non-trivial multi-unit implementation batches.
Skip if: Greenfield learners without Codex CLI, repos that forbid external exec delegation, or tasks where a single trivial edit is faster without flipping delegation flags.
When should I use this skill?
When delegation_active is true in the parent work skill and Mode Detection routes implementation through the Codex delegation workflow.
What do I get? / Deliverables
Implementation units run via codex exec in declared batch(es) while the parent agent maintains governance, or delegation drops so Claude executes locally per your choice.
- Codex-executed implementation batches per declared unit count
- Explicit delegation vs local Claude execution decision record
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Agent-tooling is the canonical shelf because the skill wires orchestrator versus implementer roles during active product construction. Codex exec delegation is tooling for how your agent stack executes code changes, not a one-off ship checklist.
Where it fits
After a written plan, delegate N backend units to Codex in one batch while you stay on git and review.
Multi-batch UI work: auto mode announces X batches before codex exec runs.
Post-delegation you re-run review locally because orchestrator never gave up merge governance.
How it compares
Orchestration slice for Codex exec handoff—not a standalone planner; use with the parent ce-work skill rather than chat-only implementation.
Common Questions / FAQ
Who is ce-work-beta for?
Indie developers using the compound-engineering plugin who actively toggle Claude-orchestrated work with Codex-backed implementation.
When should I use ce-work-beta?
During Build (agent-tooling) when delegation_active is on; after planning when batches of units should go to Codex; before Ship review when you need delegated code landed then re-reviewed locally.
Is ce-work-beta safe to install?
Delegation runs external Codex exec against your codebase—review the Security Audits panel on this page and confirm you trust Codex execution in your environment.
Workflow Chain
Then invoke: ce resolve pr feedback, ce test browser
SKILL.md
READMESKILL.md - Ce Work Beta
# Codex Delegation Workflow When `delegation_active` is true, code implementation is delegated to the Codex CLI (`codex exec`) instead of being implemented directly. The orchestrating Claude Code agent retains control of planning, review, git operations, and orchestration. ## Delegation Decision If `work_delegate_decision` is `ask`, present the recommendation and wait for the user's choice before proceeding. **When recommending Codex delegation:** > "Codex delegation active. [N] implementation units -- delegating in one batch." > 1. Delegate to Codex *(recommended)* > 2. Execute with Claude Code instead **When recommending Codex delegation, multiple batches:** > "Codex delegation active. [N] implementation units -- delegating in [X] batches." > 1. Delegate to Codex *(recommended)* > 2. Execute with Claude Code instead **When recommending Claude Code (all units are trivial):** > "Codex delegation active, but these are small changes where the cost of delegating outweighs having Claude Code do them." > 1. Execute with Claude Code *(recommended)* > 2. Delegate to Codex anyway If the user chooses the delegation option, proceed to Pre-Delegation Checks below. If the user chooses the Claude Code option, set `delegation_active` to false and return to standard execution in the parent skill. If `work_delegate_decision` is `auto` (the default), state the execution plan in one line and proceed without waiting: "Codex delegation active. Delegating [N] units in [X] batch(es)." If all units are trivial, set `delegation_active` to false and proceed: "Codex delegation active. All units are trivial -- executing with Claude Code." ## Pre-Delegation Checks Run these checks **once before the first batch**. If any check fails, fall back to standard mode for the remainder of the plan execution. Do not re-run on subsequent batches. **0. Platform Gate** Codex delegation is only supported when the orchestrating agent is running in Claude Code. If the current session is Codex, Gemini CLI, OpenCode, or any other platform, set `delegation_active` to false and proceed in standard mode. **1. Environment Guard** Check whether the current agent is already running inside a Codex sandbox: ```bash if [ -n "$CODEX_SANDBOX" ] || [ -n "$CODEX_SESSION_ID" ]; then echo "inside_sandbox=true" else echo "inside_sandbox=false" fi ``` If `inside_sandbox` is true, delegation would recurse or fail. - If `delegation_source` is `argument`: emit "Already inside Codex sandbox -- using standard mode." and set `delegation_active` to false. - If `delegation_source` is `config` or `default`: set `delegation_active` to false silently. **2. Availability Check** **Codex CLI path (pre-resolved):** !`command -v codex 2>/dev/null || true` If the line above shows an absolute path (starts with `/`, e.g., `/opt/homebrew/bin/codex`), the Codex CLI is available — proceed to the next check. Otherwise — empty, an unresolved command string like `command -v codex 2>/dev/null` left in place by a non-Claude harness that doesn't process `!` pre-resolution, or any other non-path value — run `command -v codex` via the shell/Bash tool to verify at runtime. If that prints an absolute path, the Codex CLI is available; proceed. If it fails or prints nothing, emit "Codex CLI not found (install via `npm install -g @openai/codex` or `brew install codex`) -- using standard mode." and set `delegation_active` to false. **3. Consent Flow** If `consent_granted` is not true (from config `work_delegate_consent`): Present a one-time consent warning using the platform's blocking question tool (`AskUserQuestion` in Claude Code, `request_user_input` in Codex, `ask_user` in Gemini, `ask_user` in Pi (requires the `pi-ask-user` extension)). The consent warning explains: - Delegation sends implementation units to `codex exec` as a structured prompt - **yolo mode** (`--dangerously-bypass-approvals-and-sandbox`): Full system access including network. Required for verification steps that run test