orchestrate

orchestrate is a command for coding agents from odysseus0/claude-orchestrate. It costs 50 tokens per session (549 once invoked), scanned A, original, MIT.

A command for breaking a large coding task into smaller tasks that separate workers can handle and check. It runs those workers in isolated copies of the project and combines their results.

In plain words
What is it for?
Use it to plan, parallelize, monitor, retry, and reconcile complex repository changes across multiple workers.
Why use it?
It reduces the risk of losing track of dependencies, unfinished work, or failed subtasks when a project is too large for one working session.

Command

Part of the orchestrate plugin — 1 skill, 1 command shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/odysseus0/claude-orchestrate/orchestrate
Clone the repo
git clone --depth 1 https://github.com/odysseus0/claude-orchestrate

Or install orchestrate, the plugin that ships this one along with the rest of its 1 skill, 1 command.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for orchestrate

README.md
[![agentmods](https://agentmods.dev/badge/commands/odysseus0/claude-orchestrate/orchestrate.svg)](https://agentmods.dev/commands/odysseus0/claude-orchestrate/orchestrate)
Your own site
<a href="https://agentmods.dev/commands/odysseus0/claude-orchestrate/orchestrate"><img src="https://agentmods.dev/badge/commands/odysseus0/claude-orchestrate/orchestrate.svg" alt="Measured on agentmods" height="20"></a>
Per session 50 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 549 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00050 $0.00549
Opus 5 $0.00025 $0.00275
Sonnet 5 $0.00010 $0.00110
Haiku 4.5 $0.00005 $0.00055

Measured 4d ago against content hash d4b2a81157c2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

orchestrate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/orchestrate/commands/orchestrate.md · 38 lines

What it actually says

You are the root planner for: $ARGUMENTS

Read .claude/skills/orchestrate/SKILL.md and follow it.

Discover here before you publish tasks (read existing repo state, scan for relevant code). Workers and subplanners spawn with no shared memory of what you saw — bake the essentials into their spawn prompts.

Loop hygiene:

  • Run the reconcile loop in-process inside this turn. Do not detach. The loop IS your turn.
  • After publishing tasks and surfacing the plan for user approval, enter the reconcile loop: spawn ready tasks (all task types via Bash(claude-bg-wait <worktree-name> <prompt>)), wait for handoffs, persist state, repeat until no pending task is reachable.
  • On any task transitioning to error: read the synthetic failure handoff at <workspace>/handoffs/<task>-failure.md, decide retry / split / escalate / drop, then continue. Each retry costs another spawn — after 2 attempts on the same task, prefer abandon (drop from plan.json, replan around it) over a 3rd attempt unless you have specific evidence the next retry will succeed.
  • Don't end your turn while this workspace has non-terminal tasks.
  • Auto-mode safe by construction: spawn prompts are inlined into the Bash command via single-quoted heredocs (see skills/orchestrate/references/spawning.md), so the classifier sees the full prompt content inline and doesn't have to gate spawns on opaque $(cat <file>) substitutions. Don't fall back to file-based $(cat) patterns — they're slower (extra friction) without being any safer.

After publishing the plan, surface this to the user verbatim so they know the pause-control:

This run is now driving. If you want to halt new spawns mid-run (e.g.
you spot bad upstream output or a runaway decomposition), open another
terminal and run:

    touch .orchestrate/<rootSlug>/andon

I'll see it before the next spawn cycle, pause new spawns, and wait.
In-flight work finishes; nothing new starts. To resume:

    rm .orchestrate/<rootSlug>/andon

This is the Andon cord — pause-not-kill. It works whether you're
watching this terminal or away.

(Substitute the actual rootSlug into the snippet.)

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 38 lines · 50 tokens per session scan A d4b2a81157c2

Subscribe to this mod's changes

orchestrate is a command published in the GitHub repository odysseus0/claude-orchestrate (1 stars, last pushed 3mo ago), licensed MIT. It adds 50 tokens to every session and 549 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.