avo AGENTS.md

avo AGENTS.md is an instructions file for Codex, OpenCode from gatordevin/avo. It costs 1,178 tokens per session, scanned A, original, Apache-2.0.

Repository instructions for coding agents working on avo, including normal development and operating its variation loop. A variation loop is a repeated process that runs alternatives and evaluates their results.

In plain words
What is it for?
Use it when developing avo, running its tests or doctor command, adding targets or backends, or operating and reviewing an avo run.
Why use it?
It gives agents rules for setup, testing, dependencies, prompts, and honest score handling so they do not confuse framework changes with successful runs.

Instructions file for CodexOpenCode

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/gatordevin/avo/agents-md
Clone the repo
git clone --depth 1 https://github.com/gatordevin/avo

Made for: Codex, OpenCode.

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 avo AGENTS.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/gatordevin/avo/agents-md.svg)](https://agentmods.dev/instructions/gatordevin/avo/agents-md)
Your own site
<a href="https://agentmods.dev/instructions/gatordevin/avo/agents-md"><img src="https://agentmods.dev/badge/instructions/gatordevin/avo/agents-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,178 This file is loaded in full into every session.
When invoked 1,178 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 $0.01178 $0.01178
Opus 5 $0.00589 $0.00589
Sonnet 5 $0.00236 $0.00236
Haiku 4.5 $0.00118 $0.00118

Measured 3d ago against content hash f2fc7a7bd74b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

avo 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 3d 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.md · 107 lines

How it starts

The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md — instructions for coding agents working in this repo

This file is read automatically by Codex CLI and other agents that follow the AGENTS.md convention. Claude Code users additionally have a /avo skill in .claude/skills/avo/.

There are two distinct things an agent does here, and they need different rules.


A. Working on the AVO codebase

Ordinary software work: fixing bugs, adding targets, adding backends.

python3 -m venv --system-site-packages .venv    # if Python is externally managed
.venv/bin/pip install -e ".[dev]"
.venv/bin/pytest -q                             # 48 tests, no model or network needed
.venv/bin/avo doctor

Conventions:

  • No mandatory dependencies beyond PyYAML. Anything else goes in an extra and is imported lazily, so avo doctor can report what is missing rather than the import failing.
  • The framework never calls a model. Session mode is the default precisely because it costs nothing. Keep it that way.
  • Prompts are the framework/agent interface. src/avo/prompts.py changes behaviour more than most code does. If you edit it, say in the commit message what you expect the operator to do differently.
  • Test anything touching the commit policy, the correctness gate, or the lineage. Those three are what make results trustworthy.
  • docs/PAPER_MAP.md maps every section of the paper to the code. If you change something it describes, update it.

B. Acting as the variation operator

This is the interesting one. You are the Agent in Vary(P_t) = Agent(P_t, K, f): AVO owns the lineage, the scoring, and the commit policy; you own the thinking.

Use the avo_* MCP tools if they are available, otherwise the CLI — the operations are identical. Full protocol in docs/DRIVING.md.

The loop

  1. avo_start_run / avo start --target <name> --max-steps N
  2. avo_next_step / avo prompt — read it properly, it changes every step
  3. Edit files under work/; read kb/ when you need it
  4. avo_evaluate / ./avo-eval — as often as you like, it does not touch the lineage
  5. avo_submit -m "what changed and its measured effect"
  6. On a reported stall: avo_supervisor_brief, answer it yourself, then avo_record_supervisor
  7. Repeat, then avo_plot

Read the full file on GitHub · 107 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. 3d ago First seen · 107 lines · 1,178 tokens per session scan A f2fc7a7bd74b

Subscribe to this mod's changes

avo AGENTS.md is an instructions file published in the GitHub repository gatordevin/avo (6 stars, last pushed 12d ago), licensed Apache-2.0. It adds 1,178 tokens to every session, about $0.0059 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.