What is AI_orchestrator CLAUDE.md?

A project guide for Claude Code, an AI coding assistant, describing the repository’s documentation, task queue, commands, and tests. The project is a Python task orchestrator that routes work between several AI coding tools.

It is for updating the project documentation, choosing its task-running commands, diagnosing setup, and running all or selected tests.

It gives the assistant the project rules and commands needed to make changes consistently and check that they work.

How swDomass/AI_orchestrator configures Claude Code and Codex

Reuse

Borrowing it

Nothing to install: this file belongs to swDomass/AI_orchestrator. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/swDomass/AI_orchestrator/master/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/swDomass/AI_orchestrator

Made for: Claude Code, Codex.

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/swdomass/ai_orchestrator/claude-md/github.svg)](https://agentmods.dev/instructions/swdomass/ai_orchestrator/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/swdomass/ai_orchestrator/claude-md"><img src="https://agentmods.dev/badge/instructions/swdomass/ai_orchestrator/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/swdomass/ai_orchestrator/claude-md"><img src="https://agentmods.dev/badge/instructions/swdomass/ai_orchestrator/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 20,174 This file is loaded in full into every session.

When invoked 20,174 The same file — it is already loaded in full.

Security scan D 4 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.

Model	Per session	Once invoked
Fable 5.1	$0.20174	$0.20174
Opus 5	$0.10087	$0.10087
Sonnet 5	$0.04035	$0.04035
Haiku 4.5	$0.02017	$0.02017

Measured 2d ago against content hash e732a02b612e, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade D, and why

AI_orchestrator CLAUDE.md scanned grade D with 4 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- **`scripts/safety_hook.py`** — Claude Code `PreToolUse` hook, hard-deny via `SAFETY_DENY_PATTERNS`. **The output shape decides whether it blocks at all:** Claude Code honours `decision: "block"` (legacy) and `hookSpeci

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

- **Blocked categories**: `rm -rf`, `git push --force`, `git reset --hard`, `git clean -f`, `DROP/TRUNCATE TABLE`, `DELETE FROM` without WHERE, `format`/`mkfs`/`diskpart`, fork bombs, raw disk writes, credential exfiltra

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- **`gemini.py`** — **dual-mode** (consumer Gemini CLI shut down 2026-06-18). `GEMINI_API_KEY` set → HTTP REST API via `urllib` (`generateContent`, stdlib, like `openrouter.py`); real tokens from `usageMetadata` (thinkin

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

- **`process_runner.py`** — Liveness-/Hang-Watchdog: `Popen` + 2 daemon-Reader-Threads (stdout/stderr gleichzeitig, sonst Pipe-Buffer-Deadlock), `idle_timeout` (Hang) + `hard_timeout` (Backstop), Tree-Kill (Windows `task

CLAUDE.md · 206 lines

How it starts

The opening of the file, as written. The whole thing — 206 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

Guidance for Claude Code (claude.ai/code) when working in this repository.

Detail-Docs: Implementierungs-Details pro Modul → docs/architecture/components.md. Patterns/Invariants + Phase B Feature Flag → docs/architecture/patterns.md. Bei Code-Änderungen beide Files (diese Datei + zugehörige Detail-Datei) aktualisieren.

Obsidian-Projektdoku

<your-vault>/01_Tasks/01_Projekte/.../AI-System-Intelligence-Orchestrator.md

Project

Autonomous task orchestrator routing work across Claude Code and Codex CLI, plus opt-in providers that never enter the fallback chain: OpenRouter (HTTP), Mistral Vibe (reviewer-only) and opencode CLI (tag-activated third external voice, Stufe 2 — a ZDR-guaranteed path for customer code, capped by its own OpenRouter key rather than uncapped like the other two). Gemini (HTTP API or legacy CLI) is still implemented but retired from every active path (2026-08-15). Tasks come from an Obsidian vault Markdown queue (99_System/AI/agent-queue.md). Pure Python stdlib + pyyaml, Windows-first.

Commands

# Run all tests (2447 passed / 0 failed — re-measured twice on 2026-09-05: 123 s
# and 102 s on the same tree, against ~90 s on 2026-09-04. Treat 90-125 s as the
# normal band; the spread is machine load, not a regression signal)
# -p no:randomly is load-bearing: tests/test_telegram_listener.py is
# order-dependent, so a random seed can turn the suite red with no code change.
python -m pytest tests/ -q -p no:randomly

# Run a single test file / single test
python -m pytest tests/test_parallel_runner.py -v
python -m pytest tests/test_queue_manager_regressions.py::test_extract_cwd_supports_spaces -v

# Install dependencies
pip install -r requirements.txt

# Validate setup
python orchestrator.py --doctor
python orchestrator.py --doctor --fix --yes

# Modes
python orchestrator.py                # single-shot
python orchestrator.py --watch        # continuous + heartbeat
python orchestrator.py --dry-run      # parse queue, no execute
python orchestrator.py --check-limits # provider capacity
python orchestrator.py --list-tools   # available #tool: handlers
python orchestrator.py --dashboard    # analytics web dashboard
python orchestrator.py --lint-queue   # validate agent-queue.md

# Lint / typecheck (config in pyproject.toml, no CI gate — advisory only)
ruff check .            # 1403 findings, re-measured 2026-09-05 (ruff 0.16.1); the
                        # 2026-09-02 baseline of 1351 predates providers/opencode.py,
                        # openrouter_budget.py and their tests
python -m mypy .        # 132 errors in 37 files (mypy 2.3.1, lenient config; 2026-09-04, +1/+1 from Stufe 2)

Read the full file on GitHub · 206 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.

2d ago Changed · +19 lines · +7,925 tokens per session e732a02b612e
5d ago Changed · +29 lines · +5,749 tokens per session scan C → D a269de6ee5de
9d ago First seen · 158 lines · 6,500 tokens per session scan C 3872b38cb75e

Subscribe to this mod's changes

AI_orchestrator CLAUDE.md is an instructions file published in the GitHub repository swDomass/AI_orchestrator (5 stars, last pushed 3d ago), licensed MIT. It adds 20,174 tokens to every session, about $0.1009 per session on Opus 5. A static security scan graded it D with 4 findings (reads agent configuration directories, recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Claude Code LLM automation codex gemini obsidian

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

AI_orchestrator: Instructions file for Claude Code