dev-challenge AGENTS.md

Repository instructions for a programming challenge project, with two modes: mentoring someone who is learning, or improving the challenge itself. It also defines which files are the source and which tool-specific copies must stay synchronized.

In plain words
What is it for?
Use it to decide how to respond to challenge work, mentor or implement according to the user’s request, and update the canonical instructions and their Claude, Copilot, and Cursor mirrors.
Why use it?
It prevents an agent from solving a learning exercise when guidance is expected, and helps keep duplicated instructions consistent.

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/micheltlutz/dev-challenge/agents-md
Clone the repo
git clone --depth 1 https://github.com/micheltlutz/dev-challenge

Made for: Codex, OpenCode.

Per session 1,862 This file is loaded in full into every session.
When invoked 1,862 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 $0.01862 $0.01862
Opus 5 $0.00931 $0.00931
Sonnet 5 $0.00372 $0.00372
Haiku 4.5 $0.00186 $0.00186

Measured 2d ago against content hash 2fd22a3a6bf3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dev-challenge 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 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.

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 · 183 lines

How it starts

The opening of the file, as written. The whole thing — 183 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent instructions

Guidance for coding agents (Claude Code, GitHub Copilot, Cursor, Codex, …) working in or against this repository. This file is the single source; the tool-specific files mirror it:

If you change something here that others must know, update the mirrors too.


First: which job are you doing?

The two are almost opposites. Work out which one applies before you write anything.

Job A — helping someone do a challenge

They are learning. The code is the medium, not the deliverable.

Default posture: mentor, not solver. Plan with them, review what they write, explain which criterion is failing and why. Do not produce the implementation unless they ask for it.

This is not a hard refusal. It is a default, and the user can turn it off: "just write it", "show me a reference solution", "I'm short on time" all switch you to solver mode immediately. Say what you are doing when you switch. Never be coy or withhold while pretending you cannot help — that is worse than useless.

Concretely, in mentor mode:

  • Read the spec with them and surface anything ambiguous.
  • Ask what their approach is before offering yours.
  • When a criterion fails, quote it and ask what they think is happening.
  • Review their diff against the acceptance criteria, naming the ID.
  • Offer the smallest hint that unblocks, then stop.
  • Write scaffolding, config and boilerplate freely — nobody learns from a tsconfig.json. Reserve the mentoring for the logic the challenge is about.

Job B — improving the lab itself

Normal engineering work. Read CONTRIBUTING.md and behave like any other contributor. The one non-obvious rule is the spec layer, below.


The spec layer is the source of truth

Read the full file on GitHub · 183 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 · 183 lines · 1,862 tokens per session scan A 2fd22a3a6bf3

Subscribe to this mod's changes

dev-challenge AGENTS.md is an instructions file published in the GitHub repository micheltlutz/dev-challenge (32 stars, last pushed 15d ago), licensed MIT. It adds 1,862 tokens to every session, about $0.0093 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.