llm-council CLAUDE.md

llm-council CLAUDE.md is an instructions file for coding agents from amiable-dev/llm-council. It costs 10,730 tokens per session, scanned A, original, MIT.

Project-specific instructions for working on LLM Council, a Python system where several language models discuss and review answers before producing a result.

In plain words
What is it for?
Use it when changing, running, or understanding the LLM Council package, its command-line tool, web API, or MCP server.
Why use it?
It gives coding agents the project's current architecture, commands, and design decisions, reducing mistakes caused by outdated documentation or guessed structure.

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/amiable-dev/llm-council/claude-md
Clone the repo
git clone --depth 1 https://github.com/amiable-dev/llm-council

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.

agentmods badge for llm-council CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/amiable-dev/llm-council/claude-md.svg)](https://agentmods.dev/instructions/amiable-dev/llm-council/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/amiable-dev/llm-council/claude-md"><img src="https://agentmods.dev/badge/instructions/amiable-dev/llm-council/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 10,730 This file is loaded in full into every session.
When invoked 10,730 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 $0.10730 $0.10730
Opus 5 $0.05365 $0.05365
Sonnet 5 $0.02146 $0.02146
Haiku 4.5 $0.01073 $0.01073

Measured 5d ago against content hash 85fef0ce2f8a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

llm-council 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 5d 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.md · 204 lines

How it starts

The opening of the file, as written. The whole thing — 204 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — Technical Notes for LLM Council

Durable architecture, design rationale, and gotchas for future sessions. This file is a map and a rationale, not an API reference — for function signatures, read the source (they go stale here). When you change behavior that contradicts a note below, update the note.

Project Overview

LLM Council is a 3-stage deliberation system where multiple LLMs collaboratively answer a question. The core innovation: anonymized peer review in Stage 2 (models rank "Response A/B/C…", not each other by name) so they can't play favorites.

It is a single Python package (src/llm_council/) exposing a CLI, an HTTP/REST server, and an MCP server. There is no separate frontend or backend/ directory — older docs referencing backend.main, frontend/src/, main.py, or storage.py describe a retired layout.

Running & Developing

  • Install / setup: make setup (deps + .env), or make install (deps only).
  • Run the HTTP API: llm-council serve [--host H] [--port 8000] (default port 8000). Entry point is llm_council.cli:main; the FastAPI app lives in src/llm_council/http_server.py.
  • Other CLI subcommands: setup-key (store API key in keychain, ADR-013), bias-report (cross-session bias analysis, ADR-018), install-skills, gate, ignore (inspect the verify file-selection trust boundary — --print-defaults/--explain/--init, ADR-053).
  • Quality gates: make test / make test-fast / make test-cov, make lint (ruff), make typecheck (mypy), make check (lint+typecheck), make fix. ~120 test files in tests/.
  • tests/test_openrouter.py verifies OpenRouter connectivity and that a model identifier resolves before you add it to the council.
  • Relative imports: modules import siblings as from .unified_config import …. Keep imports relative within the package.
  • ADRs live in docs/adr/. Most subsystems below cite the ADR that introduced them — read it for design context.
  • Definition of Done includes the published docs site (docs/ + mkdocs.yml nav), not just README/CLAUDE.md/CHANGELOG — the site froze at ADR-038 when it was left out of epic DoDs. tests/test_docs_drift.py enforces the mechanical parts (env reference, ADR nav, guide snippets).

Read the full file on GitHub · 204 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. 5d ago First seen · 204 lines · 10,730 tokens per session scan A 85fef0ce2f8a

Subscribe to this mod's changes

llm-council CLAUDE.md is an instructions file published in the GitHub repository amiable-dev/llm-council (41 stars, last pushed 3d ago), licensed MIT. It adds 10,730 tokens to every session, about $0.0537 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.