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.
npx agentmods add instructions/avyayalaya/agent-council/agents-mdgit clone --depth 1 https://github.com/Avyayalaya/agent-councilWrote 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.
[](https://agentmods.dev/instructions/avyayalaya/agent-council/agents-md)<a href="https://agentmods.dev/instructions/avyayalaya/agent-council/agents-md"><img src="https://agentmods.dev/badge/instructions/avyayalaya/agent-council/agents-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.02230 | $0.02230 |
| Opus 5 | $0.01115 | $0.01115 |
| Sonnet 5 | $0.00446 | $0.00446 |
| Haiku 4.5 | $0.00223 | $0.00223 |
Grade A, and why
agent-council AGENTS.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 6d 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.
How it starts
The opening of the file, as written. The whole thing — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Agent Council capability manifest
This file declares the Council to other AI systems and orchestrators. Three layers: identity, evaluability, composability. They are all in this file.
Identity
Name: agent_council (Python package); agent-council (CLI binary).
Version: 0.1.0 (P0 — Week 1 scaffold).
One-paragraph capability:
The Agent Council is a runtime-portable quality gate. Given any text artifact and a config, it runs five role-conditioned LLM deliberators in a 2-round async protocol with cross-read rebuttal, then synthesizes a single verdict — SHIP, REVISE, or HOLD — plus a revision brief and a full audit transcript. It depends on no SDK and no API keys: a runtime adapter shells out to whatever CLI is configured (
claude,gh copilot,ollama, etc.). It is designed to gate tier-1 artifacts — external-facing, irreversible, identity-shaping, or memory writes — before they persist or publish.
What it does well:
- Catches voice violations, evidence gaps, and strategic mis-alignment that the artifact's producing agent missed.
- Surfaces dissent transparently — every deliberator's score, blocking flag, and irreducible flag is logged.
- Compounds across runs — prior verdicts on the same
artifact_typeare loaded into the Adjudicator's context. - Stays portable — adding a new LLM CLI is one new file in
runtimes/.
What it is not:
- It is not a writing tool. It does not produce artifacts; it gates them.
- It is not a generic LLM-as-judge. It is structured around five named roles, each with a Context Verification Gate and a typed output schema.
- It is not a fact-checker for the open web. Source verification stays inside the Evidence deliberator's tiered framework — accuracy is the operator's responsibility.
- It is not a replacement for the operator's own review. It is the gate before the operator's review, not after.
When to invoke:
- An artifact is tier-1 (external, irreversible, identity-shaping, or a memory write).
- The artifact has cleared its producing agent's quality gate but has not yet been published / committed / persisted.
- The operator wants structured dissent before shipping — not just "looks good."
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.
- 6d ago First seen · 181 lines · 2,230 tokens per session scan A 764f73c86ad4
agent-council AGENTS.md is an instructions file published in the GitHub repository Avyayalaya/agent-council (10 stars, last pushed 1mo ago), licensed MIT. It adds 2,230 tokens to every session, about $0.0112 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 2026-08-31.
Other instructions, from other repositories
agentscope copilot-instructions.md
Copilot instructions for agentscope-ai/agentscope, covering agentscope code review guide, 1. code quality, 2. [must] code security, 3. [must] testing & dependencies and 4. code standards.
CoCoWork CLAUDE.md
Claude Code instructions for NUGURI-7/CoCoWork, covering global user preferences, 代码分工（谁动文件）, 给代码的方式（不给孤立函数）—— 只适用于后端, cocowork — claude code 指南 and 项目结构.
MassGen AGENTS.md
AGENTS.md instructions for massgen/MassGen, covering claude.md, openspec instructions, planning, instruction file parity and debugging assumptions.
MassGen CLAUDE.md
Claude Code instructions for massgen/MassGen, covering claude.md, openspec instructions, planning, instruction file parity and debugging assumptions.
hoop CLAUDE.md
Claude Code instructions for hoophq/hoop, covering claude.md, project overview, toolchain & prerequisites, architecture and module breakdown.
commonly CLAUDE.md
Claude Code instructions for Team-Commonly/commonly, covering claude.md / agents.md, 🧠 product vision & architecture philosophy, what commonly is, the architecture model and key concepts.