ardur AGENTS.md

The main instruction guide for coding agents in the ArdurAI/ardur repository. It explains the required startup steps, repository structure, commands, and trust boundaries.

In plain words
What is it for?
Use it when an agent starts work in the repository, especially to run the bootstrap script and read the generated project context.
Why use it?
It gives agents a consistent way to prepare before editing code and tells them which project information to trust.

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/ardurai/ardur/agents-md
Clone the repo
git clone --depth 1 https://github.com/ArdurAI/ardur

Made for: Codex, OpenCode.

Per session 6,115 This file is loaded in full into every session.
When invoked 6,115 The same file — it is already loaded in full.
Security scan B 1 finding. 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 $0.06115 $0.06115
Opus 5 $0.03057 $0.03057
Sonnet 5 $0.01223 $0.01223
Haiku 4.5 $0.00611 $0.00611

Measured 2d ago against content hash 0d86b0b1882a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

ardur AGENTS.md scanned grade B with 1 finding 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.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

sudo /tmp/ardur-seccomp-smoke --daemon-bin /tmp/ardur-kernelcaptured --shim-bin /tmp/ardur-exec-shim
AGENTS.md · 464 lines

How it starts

The opening of the file, as written. The whole thing — 464 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Ardur Agent Instructions

The canonical entry point for coding agents working in this repository. These instructions are mandatory. Human contributors should read CONTRIBUTING.md and docs/engineering-standards.md; this file restates the parts an agent gets wrong most often and adds the parts that only matter to agents.

0. First Action In Every New Session

Run the Conductor bootstrap before doing task-specific work:

./scripts/conductor-bootstrap.sh

Then read .context/ARDUR_CONTEXT.md. Its Generated Graph section is the authority for graph availability:

  • When the status is available, read .context/ardur-graph.md and use .context/ardur-graph.json as the machine-readable map of the repo.
  • When the status is unavailable, continue with the live source and workflow files listed in the context. Missing graph artifacts are optional in this path and are not a bootstrap failure.

If the bootstrap fails, stop and inspect the failure before editing files. A failed bootstrap usually means the local toolchain, branch state, or generated context is not trustworthy yet.

Repo truth hierarchy. Use live repo state over stale prose, in this order:

  1. git status, branch refs, and the actual files in this checkout.
  2. .github/workflows/ for the current CI surface.
  3. README.md, STATUS.md, docs/TESTING.md, docs/engineering-standards.md, docs/known-limitations.md, and docs/decisions/.
  4. Prior notes and generated .context/ files, after checking their timestamp.

If two sources conflict, cite the conflict and verify from the current tree. This repo has changed quickly around Python and Go CI; the live workflow files are the authority for what currently runs. That includes this file — if AGENTS.md disagrees with a workflow, the workflow is right and this file is a bug.

1. What Ardur Is, And What It Does Not Claim

Ardur governs AI-agent tool calls that pass through a configured adapter or proxy. It checks mission, resource, budget, and delegation constraints before that integration dispatches the call, then emits an issuer-signed, hash-linked receipt for the decision. The goal is to prove what your agents do, not just what they say.

Read the full file on GitHub · 464 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. 2d ago First seen · 464 lines · 6,115 tokens per session scan B 0d86b0b1882a

Subscribe to this mod's changes

ardur AGENTS.md is an instructions file published in the GitHub repository ArdurAI/ardur (2 stars, last pushed 3d ago), licensed MIT. It adds 6,115 tokens to every session, about $0.0306 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.