agent-orchestrator CLAUDE.md

Repository instructions for agent-orchestrator, an application that runs and supervises coding agents. They define where the app may store its state and require new interface work to follow the project's existing visual design.

In plain words
What is it for?
Useful when changing state handling, the desktop app, Electron settings, worktrees, or the web interface and its components.
Why use it?
They help prevent data from being written to the wrong location and keep interface changes consistent with the application's design rules.

Instructions file

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 instructions/untrivial-ai/agent-orchestrator/claude-md
Clone the repo
git clone --depth 1 https://github.com/Untrivial-ai/agent-orchestrator
Per session 486 This file is loaded in full into every session.
When invoked 486 The same file — it is already loaded in full.
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 $0.00486 $0.00486
Opus 5 $0.00243 $0.00243
Sonnet 5 $0.00097 $0.00097
Haiku 4.5 $0.00049 $0.00049

Measured yesterday against content hash f0f137307d48, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-orchestrator CLAUDE.md 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 yesterday.

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.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

CLAUDE.md · 35 lines

What it actually says

CLAUDE.md

Read and follow AGENTS.md for repository layout, commands, coding conventions, and hard rules.

App state lives under ~/.ao only

All app state, the daemon's data dir, running.json, worktrees, and the Electron supervisor's userData (Chromium cache, cookies, local/session storage, crash dumps), must resolve under ~/.ao (overridable via AO_DATA_DIR/AO_RUN_FILE). Never write to or read from ~/Library/Application Support or any other OS-default app-data location. frontend/src/main.ts pins Electron's userData to ~/.ao/electron; do not remove that override. See the hard rule in AGENTS.md.

Design System

Always read DESIGN.md before making any visual or UI decision — start with the "clone agent-orchestrator verbatim" banner at the top, which governs the current look.

The renderer clones the agent-orchestrator web app verbatim (~/Projects/agent-orchestrator/packages/web/src) in looks and design, with a refined-blue accent and the terminal keeping its own palette. This supersedes the older design-reference framing in DESIGN.md (per explicit user decision 2026-06-10). Build new UI from shadcn primitives (components/ui/*) where a component fits. Do not deviate without explicit user approval. In QA/review, flag any renderer code that diverges from agent-orchestrator — do not re-flag old design-reference mismatches.

When showing or demoing frontend changes, run ao preview [url] from inside the session so the change renders in the desktop browser panel (the inspector rail's Browser tab); do not just describe it. ao preview updates the panel non-disruptively — it does not steal focus or force the Browser tab open if the user is looking at something else, only badging it as unseen. If the Browser tab isn't already the one the user has open, say so in your reply (e.g. "check the Browser tab") so the change doesn't go unnoticed behind the badge.

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. yesterday First seen · 35 lines · 486 tokens per session scan A f0f137307d48

Subscribe to this mod's changes

agent-orchestrator CLAUDE.md is an instructions file published in the GitHub repository Untrivial-ai/agent-orchestrator (10,763 stars, last pushed today), licensed Apache-2.0. It adds 486 tokens to every session, about $0.0024 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.