mcpField
P
pi-delegate

pi-delegate

github.com/amelnagdyAgent workflows2.5k3.6k installsUpdated 2w ago
Repository

About pi-delegate

pi-delegate is a skill for orchestrators that assigns bounded coding tasks to the Pi coding agent CLI, then supports review and committing by the orchestrator. It uses a brief and relay script, with read-only controls and explicit handling of project trust.

Key features

  1. 01Delegate bounded tasks to the Pi coding agent CLI
  2. 02Dispatch briefs through a bundled relay
  3. 03Resume Pi sessions with a delta brief
  4. 04Restrict callable tools in read-only mode
  5. 05Review diffs and rerun project checks

Use cases

  1. 01Implement a scoped repository change
  2. 02Request read-only code review or diagnosis
  3. 03Resume implementation after review identifies needed changes

MCP support

For Codex background orchestration, the skill requires the separate codex-background utility and its registered delegate_run MCP tool; it does not silently fall back to shell polling. Other orchestrators use the shell workflow.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/amelnagdy/delegate-skills --skill pi-delegate

After installing, type/pi-delegatein Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

references/
dispatch-and-poll.md
multi-task-queues.md
review-and-land.md
writing-the-brief.md
scripts/
relay.mjs

Details

From the project README

Pi Delegate

You are the orchestrator. Delegate a bounded coding task to a separate implementer - the Pi coding agent CLI - then review what it produced and land it yourself. You write the brief and own the judgment; the implementer makes changes in its own session; you verify and commit.

The loop needs only a shell command and file access, so any comparable orchestrator can drive it.

When NOT to use this

  • The task is small enough to do inline; delegation overhead is not worth it.
  • The pi CLI is not installed or authenticated.
  • You need a sandboxed implementer. Pi has no sandbox and no permission modes; --read-only restricts the tool surface, but a write-capable run executes without prompts.

Prerequisites (check once)

  1. Install pi with npm install -g @earendil-works/pi-coding-agent.
  2. Authenticate: /login inside pi for a subscription provider, or an API-key environment variable / pi's auth file for an API-key provider.
  3. Confirm pi --version succeeds.
  4. Work in, or point --cd at, the target git repository.

Choose the model (optional)

Omit --model to use pi's configured default. To pick another, choose from pi --list-models and pass an explicit id or pattern like <provider>/<model-id> or sonnet:high. The relay accepts letters, digits, and . _ : / - only (the value reaches a shell on Windows), so glob patterns with * are not forwarded.

The loop

Run these five steps per task. Steps 1, 4, and 5 require judgment; 2 and 3 are mechanical.

1. Write the brief

Pi sees only the text you send plus what it can inspect in the workspace - no chat history or shared context. Include the goal, current state, what to change, what to leave untouched, the project's actual gates, and a report contract. Tell pi not to commit. Keep one task per brief. Pi auto-loads AGENTS.md/CLAUDE.md context files from the workspace and its parents, so repo instructions reach it without inlining. See references/writing-the-brief.md.

2. Dispatch

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.

Use the bundled relay. It pipes the brief to pi --mode json on stdin, captures the JSON event stream, and writes result.json. (<skill-dir> is the installed folder containing this SKILL.md.)

node "<skill-dir>/scripts/relay.mjs" --brief brief.txt --cd /path/to/repo
# choose a model:                          add --model <id from pi --list-models>
# choose a provider:                       add --provider <name>
# read-only run (review/diagnosis):        add --read-only
# trust project .pi resources:             add --approve
# resume the most recent session:          add --resume-last  (delta brief only)
# resume a specific session:               add --session <id> (delta brief only)
# hard time limit (watchdog):              add --timeout 2h  (the 30m default suits short runs; implementation briefs routinely need 1-2h)
# see all options:                         node .../relay.mjs --help

The child process's cwd pins the workspace. The relay writes artifacts under the system temp dir by default and never commits. See references/dispatch-and-poll.md.

3. Wait for completion

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 relay blocks until pi finishes. Run it with the orchestrator's background-command facility, or background it in the shell and poll for result.json. A pre-run usage error exits 2 and writes no result; a missing pi exits 127 and writes status: "pi_unavailable".

Trust process state and the working tree over a progress display. Completion means the process exited and result.json exists. Pi's full report is the finalMessage field in result.json (also printed in full on stdout between the report markers).

4. Review - do not trust the self-report

Treat pi's final message and gate claims as claims:

  • Re-run the project's gates yourself.
  • Read the diff against the brief, starting with touchedFiles.
  • Run relevant guard skills if installed.
  • Round-trip migrations and grep for dangling references after removals or renames.

See references/review-and-land.md.

5. Land it

The implementer edits the working tree; the orchestrator commits. Commit only after the gates pass and the diff holds. If rework is needed, send a delta brief with --resume-last or --session <id>, then review again.

Autonomy and permissions

Pi has no sandbox and no permission modes. A default headless run reads, writes, edits, and executes shell commands with no prompts - the controls are:

  1. --read-only restricts pi's callable tools to --tools read,grep,find,ls across built-in, extension, and custom tools. Installed extension code still runs with the user's host permissions.
  2. The relay passes --no-approve by default, so project .pi settings, extensions, and skills stay untrusted. --approve is the explicit opt-in for a repository the user trusts.
  3. touchedFiles and the diff are the record of what changed. Inspect them after every run.

Authorization model

Delegation is something the human opts into. Once they have ("run this queue", "proceed"), committing verified, gate-passing work is the agreed contract. Two limits remain: surface, don't absorb (report pi's design decisions, defensible-but-unasked turns, and non-blocking nitpicks) and stop for scope changes (if correct completion needs going beyond the brief, ask instead of expanding the mandate). See references/review-and-land.md.

References

Comments

Markdown supported