Loading
>-
The skill file stays in the publisher's repository.
Paste the copied command in a terminal. It installs this SKILL.md into that agent.
npx skills add amElnagdy/delegate-skills --skill commandcode-delegate -a cursor -y
>-
cmd) as a background implementer, then review its
diff and land it yourself. Use this whenever the user wants to hand implementation work to Command
Code — phrasings like "have Command Code do X", "delegate this to commandcode", "run it through
cmd", or "use Command Code to implement/fix/refactor" — or to run a queue of coding tasks through
Command Code while staying the reviewer. DO NOT USE for tasks small enough to do inline, or when the
user wants the code written directly without delegating.
license: MIT
compatibility: Requires the Command Code CLI (cmd, or cmdc on Windows, from commandcode.ai) installed and authenticated, Node 22+, and git. The orchestrating agent must be able to run shell commands and read files. Shell examples assume bash/zsh (macOS/Linux).
metadata:
version: 0.5.0You are the orchestrator. This skill lets you hand a bounded coding task to a separate
implementer — the Command Code CLI (cmd) — then review what it produced and land it yourself.
You write the brief and own the judgment; Command Code does the typing in your working tree; you
verify and commit.
Nothing here is specific to one orchestrating agent. The loop needs only the ability to run a shell command and read a file, so it works the same whether you are Claude Code, OpenCode with a selected model, or any comparable agent. (It is designed for and run on Claude Code; treat other orchestrators as designed-for, not yet proven.)
cmd CLI is not installed or not authenticated (run cmd login)./review).cmdc --version does not work. Upstream recommends WSL for stable Windows use.Command Code's headless mode has exactly two states, with nothing in between:
-p with no --yolo): read, grep, and glob work. Every write, edit, and shell call is
refused by the CLI's permission layer, and headless mode has no prompt to grant them mid-run. This
is the relay's --read-only.--yolo (alias --dangerously-skip-permissions): every tool is allowed, anywhere the process
can reach. There is no filesystem sandbox and no path restriction. This is what an implementation
run needs, so the relay passes it by default.--permission-mode auto-accept and --tools-all do not lift the headless write gate. Direct CLI
probes refused write, edit, and shell with both. So an implementation run
through Command Code is a full-trust run: scope it with a tight brief and a clean working tree, not
with a sandbox. The brief is guidance, and a git worktree isolates a checkout without containing the
process. If writes outside the target tree are unacceptable, use an OS-enforced sandbox such as
codex-delegate or run this one inside a container.
cmd --version succeeds and cmd status reports authenticated. If not, install Command Code and
run cmd login.command -v cmd shows the active cmd. On native
Windows, use cmdc --version; cmd is the system shell. The relay uses cmdc there and launches
its npm .cmd shim through cmd.exe. COMMANDCODE_BIN remains an absolute-path override and must
never point to the system command interpreter. The relay records the version it ran in
result.json, so a wrong binary is visible after the fact.--cd at) the target git repository, and its tree is clean before you
dispatch — a full-trust run is much easier to review against a clean baseline.Run these five steps per task. Steps 1, 4, and 5 are your judgment; 2 and 3 are mechanical.
Command Code sees only the text you send — no repo memory, no chat history, no shared context
(beyond the repo's own AGENTS.md, which it reads automatically). Everything the task needs goes in
the brief: the goal, the current state, what to change, what to leave untouched, the project's
actual gate commands (discover them from the repo's AGENTS.md/CLAUDE.md/Makefile — do not assume),
and a report contract. Tell it that it will not commit (you will). Keep one task per brief. Full
guidance and a template: references/writing-the-brief.md.
For Codex orchestration of background work, install and configure the separate
codex-background support utility first. If its MCP tools or registry are missing, stop and
report the setup requirement; do not silently fall back to model-driven polling. Then invoke its
registered delegate_run MCP tool once in place of the shell example below.
Follow that skill for structured common arguments, unchanged provider flags and reattachment by
runId. Keep the request pending until completion, then continue this skill's review and landing
steps. Other orchestrators keep the shell workflow below.
Send the brief to Command Code with the bundled helper. It wraps cmd -p, captures the run, and
writes a structured result.json — so your only job is "run a command, read a file." (<skill-dir>
below is this skill's installed directory — the folder containing this SKILL.md, i.e. the directory
you loaded the skill from. Claude Code prints it as "Base directory for this skill" when the skill
loads; on other orchestrators use that same directory — if unsure where it landed, run
find ~ -name relay.mjs -path '*commandcode-delegate*' and substitute the directory above it.)
node "<skill-dir>/scripts/relay.mjs" --brief brief.txt --cd /path/to/repo
# read-only (review/diagnosis, no edits): add --read-only
# continue the exact session: add --session <sessionId> (from result.json; send only the delta brief)
# fallback when no session id is available: add --continue-last
# hard time limit (watchdog): add --timeout 2h (default: off; implementation runs routinely need 1-2h)
# see all options: node .../relay.mjs --help
The helper defaults to a write-capable (--yolo) run, which intentionally edits the target repository.
Its temp directory keeps only relay artifacts out of that repository. The relay never commits —
see step 5. Mechanics, flags, and the
result.json shape: references/dispatch-and-poll.md.
Codex using codex-background keeps the original MCP request pending: no model-driven sleep,
status or log polling. Use delegate_wait only to reattach after an interrupted request.
The helper blocks until Command Code finishes, so back it with whatever your orchestrator offers and resume when it returns:
run_in_background: true; you are notified on completion.… & in bash/zsh, or your shell's equivalent. The run is done when result.json
exists with a status. (A pre-run usage error — bad args or an empty brief — instead exits with code
2 and a stderr message and writes no result file, so check the exit code too. A missing cmd binary
exits 127 but does write a result.json with status commandcode_unavailable.)Do not trust progress trackers over reality: a run is finished when result.json is written and the
process has exited. Read the working tree, not a status line. The implementer's full report is the
finalMessage field in result.json (also printed in full on stdout between the report markers).
result.json includes Command Code's own summary and gate claims. Re-verify, don't accept:
touchedFiles in the result is your starting point — and because the run was
full-trust, check for edits outside the paths the brief named, not just inside them.guard-skills) — this skill produces the work; those skills judge it.Full checklist: references/review-and-land.md.
The relay never commits, but it cannot stop Command Code under --yolo from writing .git. The brief
forbids implementer commits, and the reviewer compares HEAD with the recorded pre-dispatch baseline
before landing anything. The orchestrator commits. Only after the gates pass and the diff holds:
--session <sessionId> from the prior result.json
(use --continue-last only when no session id is available), and review again.The relay doubles as a clean way to get an adversarial second opinion: dispatch --read-only with a
brief that lists the agreed points, then each contested point with both positions, and ask Command
Code to defend or concede each — deliverable in its final message, touching no files. The read-only
guarantee here is the CLI's own permission layer rather than an OS sandbox, so the relay also checks
it after the fact: readOnlyViolation: false means the Git-visible detector saw no change (ignored
or outside-repository paths are not covered); true means it saw one; null means git could not tell.
Delegation is something the human opts into. Once they have ("run this queue", "proceed"), committing verified, gate-passing work is the agreed contract — that is the whole point. Two limits on that mandate: surface, don't absorb (report Command Code's design decisions, defensible-but-unasked turns, and non-blocking nitpicks rather than silently keeping them) and stop for scope changes (if correct completion needs going beyond the brief, ask — don't expand the mandate yourself). The full treatment is in references/review-and-land.md.
relay.mjs flags, the
result.json contract, backgrounding per orchestrator, and recovery when a run misbehaves.