orchestrate

A conductor skill that coordinates coding agents through a plan, assigning implementation, review, and criticism work in waves. It tracks task dependencies and handles review outcomes.

In plain words
What is it for?
Use it to execute an existing plan, break work into tasks, send independent tasks to agents, review their changes, and run final integration checks.
Why use it?
It organizes multi-step development work so tasks are assigned, checked, corrected when needed, and verified together.

Skill for Claude CodeCodex

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 skills/codecast-sh/codecast/orchestrate
Any agent
npx skills add codecast-sh/codecast --skill orchestrate
Clone the repo
git clone --depth 1 https://github.com/codecast-sh/codecast

Made for: Claude Code, Codex.

Per session 56 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,398 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 $0.00056 $0.01398
Opus 5 $0.00028 $0.00699
Sonnet 5 $0.00011 $0.00280
Haiku 4.5 $0.00006 $0.00140

Measured yesterday against content hash 2ed98decaba8, 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 yesterday.

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.

packages/cli/orchestration/skills/orchestrate/SKILL.md · 164 lines

How it starts

The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are now acting as a conductor — your job is to drive a plan to completion by coordinating worker agents, not by writing code yourself.

Core loop

1. Read plan state
2. Identify ready tasks (unblocked, unclaimed)
3. Group into a wave of independent work
4. Spawn implementer agents (max 3-5 parallel, worktree-isolated)
5. When workers complete, spawn reviewers for their output
6. Handle review verdicts (pass → merge, needs_changes → re-implement, reject → escalate)
7. If all tasks done → run integration verification + critic sweep
8. If tasks remain → goto 2

Before you start

Read the plan and ground yourself:

cast plan context <plan_id>    # or --current if already bound

If there's no plan doc yet, create one to serve as shared knowledge for all agents:

cast doc create "<Plan Title> — Shared Context" -t plan -c "Goal: ...
Constraints: ...
Decisions: (none yet)"

Decomposing (if tasks don't exist yet)

If the plan has no tasks, decompose it yourself:

  1. Read the plan goal and acceptance criteria
  2. Explore the codebase — use Glob, Grep, Read to understand the relevant code
  3. Create tasks with clear acceptance criteria, ordered by dependency:
cast task create "Add status field to tasks schema" -t feature -p high --plan <plan_id>
cast task create "Update task API to filter by status" -t feature -p high --plan <plan_id>
cast task dep <second_id> --blocked-by <first_id>

Rules for decomposition:

  • Schema/data model first, then backend logic, then UI, then polish
  • Each task should produce a testable, committable change
  • Include test-writing as part of feature tasks, not separate tasks
  • Name tasks with specific files/functions, not generic descriptions
  • Reference actual code paths you found during exploration

Spawning workers

For each ready task, spawn an implementer agent in an isolated worktree:

Agent(
  name: "impl-<task_short_id>",
  subagent_type: "implementer",
  isolation: "worktree",
  prompt: "Implement task <task_short_id>: <task_title>\n\nPlan: <plan_id>\n<task description and acceptance criteria>"
)

Read the full file on GitHub · 164 lines

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. yesterday First seen · 164 lines · 56 tokens per session scan A 2ed98decaba8

Subscribe to this mod's changes

orchestrate is a skill published in the GitHub repository codecast-sh/codecast (30 stars, last pushed yesterday), licensed MIT. It adds 56 tokens to every session and 1,398 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 2026-08-30.