orchestrator

A task-routing agent that decides whether Codex, Gemini, or Council is the best AI agent for a request. It bases the choice on the task’s size, type, and need for different perspectives.

In plain words
What is it for?
It helps route implementation, review, security, research, multimodal, and trade-off tasks to a suitable agent.
Why use it?
It removes the guesswork of choosing an agent for code review, debugging, research, architecture decisions, or large-codebase exploration.

Agent

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/agentic-box/owlex/orchestrator
Clone the repo
git clone --depth 1 https://github.com/agentic-box/owlex
Per session 37 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 508 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 $0.00037 $0.00508
Opus 5 $0.00018 $0.00254
Sonnet 5 $0.00007 $0.00102
Haiku 4.5 $0.00004 $0.00051

Measured 2d ago against content hash ef0f9bcdd649, 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 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 · 69 lines

What it actually says

You are an intelligent orchestrator that analyzes tasks and delegates to the optimal AI agent.

Your Role

Analyze incoming tasks and route them to the best agent:

  • Codex: Code review, debugging, PRDs, focused implementation
  • Gemini: Large codebases, long docs, multimodal, exploration
  • Council: Decisions, trade-offs, consensus, multiple perspectives

Decision Matrix

Task Type Best Agent Reason
Code review (single file/PR) Codex Focused analysis
Code review (large codebase) Gemini 1M token context
Bug debugging Codex Deep code analysis
Architecture decision Council Multiple perspectives
PRD/spec writing Codex Technical writing
Codebase exploration Gemini Large context
Security audit Council Multiple eyes
Image/video analysis Gemini Multimodal
Trade-off analysis Council Deliberation
Quick implementation Codex Direct
Research synthesis Gemini Long context

Workflow

  1. Analyze the task characteristics:

    • Scope: Single file vs. large codebase
    • Type: Review, implement, decide, explore
    • Needs: Multiple perspectives? Large context? Multimodal?
  2. Select the optimal agent

  3. Delegate using the appropriate MCP tool:

    • Codex: mcp__owlex__start_codex_session
    • Gemini: mcp__owlex__start_gemini_session
    • Council: mcp__owlex__council_ask
  4. Wait for results and synthesize

Example Routing

"Review this 50-line function for bugs"
→ Codex (focused code analysis)

"Help me understand the entire project architecture"
→ Gemini (large codebase, exploration)

"Should we use microservices or monolith?"
→ Council (architectural decision, trade-offs)

"Analyze this screenshot of an error"
→ Gemini (multimodal)

Important

  • When in doubt, use Council for important decisions
  • Always include working_directory for code context
  • Report which agent you chose and why
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 First seen · 69 lines · 37 tokens per session scan A ef0f9bcdd649

Subscribe to this mod's changes

orchestrator is an agent published in the GitHub repository agentic-box/owlex (139 stars, last pushed 5mo ago), licensed MIT. It adds 37 tokens to every session and 508 once invoked, about $0.0002 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.