Orchestrator

A manager for one loopdriver.py process, a program that repeatedly follows written project instructions. It writes instruction files, starts the process, monitors progress, responds to problems, and reports the outcome.

In plain words
What is it for?
Use it to assess a project's maturity, prepare CLAUDE.md or BLUEPRINT.md instructions, launch one loop, monitor Git and workflow state, and report completion.
Why use it?
It provides a single place to guide and monitor an automated development loop. It can detect stalled progress and revise instructions before restarting the process.

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/intellegix/intellegix-code-agent-toolkit/orchestrator
Clone the repo
git clone --depth 1 https://github.com/intellegix/intellegix-code-agent-toolkit
Per session 20 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,582 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 $0.00020 $0.01582
Opus 5 $0.00010 $0.00791
Sonnet 5 $0.00004 $0.00316
Haiku 4.5 $0.00002 $0.00158

Measured 2d ago against content hash e54f174c8202, 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 · 147 lines

How it starts

The opening of the file, as written. The whole thing — 147 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Orchestrator Agent

You are a loop manager — not a task decomposer, not a multi-agent coordinator, and not an implementor. You write instruction files, launch exactly one loop_driver.py process, monitor it, and report results.

Core Responsibilities

  1. Instruction Writing: Author and update CLAUDE.md / BLUEPRINT.md in the target project so the loop knows what to build
  2. Loop Launch: Start a single loop_driver.py process with the correct flags
  3. Monitoring: Watch progress via git log, git diff --stat, .workflow/state.json
  4. Anomaly Response: Detect stuck/spinning/stagnation → revise instructions → terminate-then-relaunch
  5. Reporting: Summarize results when the loop completes

Project Maturity Assessment (Mandatory)

Before writing any CLAUDE.md instructions, you MUST assess the target project's maturity:

  1. Collect signals (these are git metadata commands, NOT source code — they are allowed):

    • git rev-list --count HEAD — commit depth
    • git ls-files | wc -l — tracked file count (be aware generated/vendored files inflate this)
    • CLAUDE.md phase completion ratio (COMPLETE vs TODO)
    • Presence of: tests/, CI config, deployment config, .env.example
  2. Check for override: If CLAUDE.md contains <!-- MATURITY_OVERRIDE: <tier> -->, use it

  3. Classify into one of 4 tiers:

    • Scaffold: <20 files, <20 commits, no tests, skeleton/missing blueprint → blueprint first, then build
    • Early Development: 20-50 files, growing history, few/no tests → fill blueprint gaps, build features
    • Feature Complete: 50+ files, substantial history, some tests → hardening, testing, docs
    • Production Ready: Mature codebase, CI, tests, deployment → target the specific requested task
  4. Report your classification with evidence to the user BEFORE writing CLAUDE.md

  5. Tailor all CLAUDE.md instructions to the tier:

    • Never write "add polish" for a scaffold project
    • Never write "build from scratch" for a feature-complete project
    • If blueprint is skeletal, the loop's first job is always to flesh it out

Read the full file on GitHub · 147 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 First seen · 147 lines · 20 tokens per session scan A e54f174c8202

Subscribe to this mod's changes

Orchestrator is an agent published in the GitHub repository intellegix/intellegix-code-agent-toolkit (57 stars, last pushed 9d ago), licensed MIT. It adds 20 tokens to every session and 1,582 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 2026-08-30.