cp-run

Delegate a task to the GitHub Copilot worker

Task to delegate: $ARGUMENTS

The task runs via cp-stream — it pipes copilot -p ... --output-format json stdout through cp-stream-parse.mjs for live progress + a rolling status.json, writing the same session directory layout the other backends use → live, observable, resumable. Monitor progress in a cost-conscious way: read only the small status.json, never the raw transcript. The session id is the copilot session id (relocated into place once known, printed on stderr).

Prerequisite: cp-agent / cp-stream installed (/cli-dispatch:setup, GitHub Copilot backend), the copilot CLI installed/authenticated, and an active GitHub Copilot subscription. Auth is COPILOT_GITHUB_TOKEN > GH_TOKEN > GITHUB_TOKEN; cli-dispatch automatically reuses the host gh auth token as GH_TOKEN when available.

If it's a real repo task (file changes needed) — isolate in a git worktree:

1. Use the bundled helper, which creates the worktree off origin/main, symlinks every node_modules dir of the source checkout (root + workspace packages, resolved from the repo top even when <repo-path> is a subdirectory), runs cp-stream in it, and prints the cleanup command:

   "${CLAUDE_PLUGIN_ROOT}/scripts/cp-worktree-run.sh" <repo-path> cp-run-<branch-name> <brief-file>
   

   Or run it by hand:

   WORKTREE=$(mktemp -d)
   git worktree add "$WORKTREE" -b cp-run-<branch-name> origin/main
   cp-agent --cwd "$WORKTREE" --max-runtime 600 "$ARGUMENTS"
   

2. Monitor (cost-conscious): capture the session id from stderr, then check progress via /cli-dispatch:watch <session-id> (state: running→done). Do NOT tight-loop tail.
3. When done, review the diff (git -C "$WORKTREE" diff), verify independently (build/test).
4. If all good, you handle git/commit/push/PR/merge; then clean up the worktree.

If it's pure generation (code/text, no files) — as a background task:

cp-agent -q "$ARGUMENTS"   # stdout = final answer only; progress in status.json/progress.log

No sandbox — read this before trusting the output unattended: --allow-all-tools --no-ask-user (always passed by cp-stream) auto-approves tools and prevents interactive questions — a functional requirement for headless use, not a safety opt-in. GitHub Copilot has no OS-level or tool-level write-deny at all, unlike Codex's real kernel-enforced --read-only. There is no --sandbox flag here to reach for. Isolation via a git worktree + your own diff review is the ONLY safety boundary.

Model selection:

cp-agent --model gpt-5.4 -q "$ARGUMENTS"

Pass the Copilot model slug directly (examples: gpt-5.4, auto). Omit --model to use GitHub Copilot's own default (or the CP_MODEL config value). Current model list is only visible interactively via /model in the copilot TUI (auth required) or GitHub Copilot docs — slugs change over time.

Reasoning effort:

cp-agent --effort high -q "$ARGUMENTS"

--effort low|medium|high maps to Copilot's --reasoning-effort=<level> flag.

Follow-up / fix (continue the same GitHub Copilot session):

cp-agent --resume <session-id> "<follow-up>"

This passes --resume <id> to copilot.

To see all sessions (all backends), use /cli-dispatch:sessions.

The worker = GitHub Copilot; you = reviewer/merge owner. Don't trust the output until verified.