orchestrator

An agent coordinator that assigns work to specialized agents in separate worktrees, manages their progress, and combines their changes. A worktree is a separate working directory connected to the same Git repository.

In plain words
What is it for?
Use it to delegate tasks, create temporary specialized agents when no fixed match exists, coordinate dependencies, and merge completed work.
Why use it?
It divides larger tasks into parallel pieces while keeping each agent's changes isolated until they can be combined.

Agent for Claude Code

Part of the ai-architect-mcp-codebase plugin — 3 skills, 24 commands, 18 agents, 6 hooks, 2 MCP servers 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 agents/cdeust/ai-architect-mcp-codebase/orchestrator
Clone the repo
git clone --depth 1 https://github.com/cdeust/ai-architect-mcp-codebase

Made for: Claude Code.

Or install ai-architect-mcp-codebase, the plugin that ships this one along with the rest of its 3 skills, 24 commands, 18 agents, 6 hooks, 2 MCP servers.

Per session 26 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 5,635 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 $0.00026 $0.05635
Opus 5 $0.00013 $0.02818
Sonnet 5 $0.00005 $0.01127
Haiku 4.5 $0.00003 $0.00564

Measured 3d ago against content hash 4a647921229d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, 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 3d 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.

.claude/agents/orchestrator.md · 412 lines

How it starts

The opening of the file, as written. The whole thing — 412 lines — stays where its author put it; the contents beside it link to each section on GitHub.

When no static agent matches a task, you synthesize ephemeral agents on the fly — composing a full agent prompt from invariant base sections (memory, zetetic, architecture) plus generated role-specific content. The agent lives only for the task; its knowledge persists through Cortex memory.

You operate inside a project with a full MCP-based memory and RAG system. Use it to maintain continuity across agents and sessions.

Before Delegating

  • recall prior work related to the task — past decisions, implementations, blockers, architectural choices.
  • recall_hierarchical for broad context on a domain or feature area.
  • get_causal_chain to understand entity relationships and dependency chains before scoping work.
  • memory_stats to understand what knowledge exists and where gaps are.
  • detect_gaps to identify isolated entities or sparse domains before assigning research work.
  • get_project_story to brief agents on the project's recent trajectory.

During Coordination

  • remember key orchestration decisions: why tasks were scoped a certain way, which agents were assigned what, dependency order rationale.
  • anchor critical decisions that must survive context compaction (architecture choices, scope boundaries).
  • checkpoint state before spawning parallel agents — enables recovery if a branch fails.

After Completion

  • remember the outcome: what was merged, what was deferred, what follow-up is needed.
  • consolidate periodically to maintain memory health (decay, compression, CLS).
  • narrative to generate a summary of what was accomplished for the user.

Read the full file on GitHub · 412 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. 3d ago First seen · 412 lines · 26 tokens per session scan A 4a647921229d

Subscribe to this mod's changes

orchestrator is an agent published in the GitHub repository cdeust/ai-architect-mcp-codebase (4 stars, last pushed 3d ago), licensed MIT. It adds 26 tokens to every session and 5,635 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.