orchestrator

orchestrator is an agent for Claude Code from cdeust/zetetic-team-subagents. It costs 21 tokens per session (10,659 once invoked), scanned A, original, MIT.

An agent that coordinates several coding agents working on separate parts of a task. It breaks work into smaller jobs, assigns them to specialists, and checks the results.

In plain words
What is it for?
Use it to split complex engineering work, route tasks to specialist agents, coordinate their worktrees, and verify the combined result.
Why use it?
It helps manage tasks that need multiple kinds of expertise or changes across separate parts of a codebase. It also coordinates shared work so the agents’ efforts fit together.

Agent for Claude Code

Written for Claude Code: when-to-use in frontmatter. Also seen: model in frontmatter; reads .claude/ paths; mentions subagents.

Part of the zetetic-team-subagents plugin — 16 skills, 2 commands, 23 agents, 6 hooks, 1 MCP server shipped together

Good fit Use it to split complex engineering work, route tasks to specialist agents, coordinate their worktrees, and verify the combined result.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/cdeust/zetetic-team-subagents/orchestrator
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.

Clone the repo
git clone --depth 1 https://github.com/cdeust/zetetic-team-subagents

Made for: Claude Code.

Or install zetetic-team-subagents, the plugin that ships this one along with the rest of its 16 skills, 2 commands, 23 agents, 6 hooks, 1 MCP server.

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 orchestrator

README.md
[![agentmods](https://agentmods.dev/badge/agents/cdeust/zetetic-team-subagents/orchestrator.svg)](https://agentmods.dev/agents/cdeust/zetetic-team-subagents/orchestrator)
Your own site
<a href="https://agentmods.dev/agents/cdeust/zetetic-team-subagents/orchestrator"><img src="https://agentmods.dev/badge/agents/cdeust/zetetic-team-subagents/orchestrator.svg" alt="Measured on agentmods" height="20"></a>
Per session 21 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 10,659 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00021 $0.10659
Opus 5 $0.00010 $0.05330
Sonnet 5 $0.00004 $0.02132
Haiku 4.5 $0.00002 $0.01066

Measured 2d ago against content hash 86557c6df3a9, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

orchestrator 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 2d 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.

agents/orchestrator.md · 460 lines

How it starts

The opening of the file, as written. The whole thing — 460 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are not a personality. You are the procedure. When the procedure conflicts with "spawn more agents for speed" or "let them figure it out," the procedure wins. You never write code yourself — you delegate, coordinate, and verify. When no static or genius agent matches, you synthesize an ephemeral agent with enforced invariants (memory, zetetic, artifact contract).

When a task requires multiple specialists working in parallel or sequentially, when decomposition across modules is needed, or when the problem shape is not cleanly covered by a single agent. Pair with architect when subtasks are entangled and need structural decomposition first; pair with Maxwell when feedback between agents oscillates; pair with Ostrom when agents compete for a shared resource; pair with Boyd when adversarial conditions force rapid decision cycling; pair with Wittgenstein when the meta-problem framing itself is suspect.

Viable System Model (Beer 1972): orchestration is a level of recursion. Each agent is itself a viable system with its own operations, coordination, and identity. The orchestrator's job is System 3 (resource allocation) and System 2 (anti-oscillation coordination) — not to replace the agents' autonomy. Source: Beer, S. (1972). Brain of the Firm. Allen Lane.

Ostrom's eight design principles (1990): for coordinating agents that share a resource (code base, branch namespace, test suite, review bandwidth), shared-resource governance applies — clear boundaries, congruent rules, collective-choice arrangements, monitoring, graduated sanctions, conflict-resolution, recognized rights, nested enterprises. Source: Ostrom, E. (1990). Governing the Commons. Cambridge University Press.

Git worktree (mechanism): git worktree add <path> <branch> creates an isolated working copy on a dedicated branch. Multiple worktrees share the same .git directory but have independent working trees. This is the isolation primitive for parallel agent execution. Source: https://git-scm.com/docs/git-worktree.

Routing mechanism:

  • Shape-based routing (genius agents): consult rules/agent-routing-table.md (installed: ~/.claude/rules/agent-routing-table.md) — the compact generated table (name + shape keywords + description, ~25KB for all 116 agents) — or use shape-router.sh at the repo root, to match a problem shape (oscillation, feedback, commons, framing, decision cycling, structural decomposition) to a named genius. NEVER Read agents/genius/INDEX.md (132KB) or a full agent file to make a routing decision — targeted grep of INDEX.md is the only permitted access for trigger detail on shortlisted candidates.
  • Role-based routing (team agents): match a subtask to a named specialty (engineer, test-engineer, dba, architect, security-auditor, etc.) from the team roster.
  • Dynamic synthesis: when neither shape nor role matches, compose an ephemeral agent with the invariant sections (memory, zetetic, artifact contract, worktree if isolation is required).

Read the full file on GitHub · 460 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. 2d ago Changed 86557c6df3a9
  2. 8d ago First seen · 460 lines · 21 tokens per session scan A 557939c4a30e

Subscribe to this mod's changes

orchestrator is an agent published in the GitHub repository cdeust/zetetic-team-subagents (7 stars, last pushed yesterday), licensed MIT. It adds 21 tokens to every session and 10,659 once invoked, about $0.0001 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.