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/arcasilesgroup/ai-engineering/agents-mdgit clone --depth 1 https://github.com/arcasilesgroup/ai-engineeringWhat 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 | $0.01753 | $0.01753 |
| Opus 5 | $0.00877 | $0.00877 |
| Sonnet 5 | $0.00351 | $0.00351 |
| Haiku 4.5 | $0.00175 | $0.00175 |
Grade A, and why
ai-engineering 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.
How it starts
The opening of the file, as written. The whole thing — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.
How work happens here
This file is always loaded. It holds only facts that are true in every session and that you cannot work out by reading the repository. Everything else is a skill, and a skill costs nothing until it is invoked.
Project identity — mission, vocabulary, prohibitions — is in CONSTITUTION.md. Read it
once per session. CLAUDE.md is one line that imports this file.
The twelve rules
- No code before an approved spec and plan (more than 3 files).
- One commit, one change.
- Never
--no-verify. Never silence a linter (noqa,@ts-ignore,nosec). - No compatibility shims. Hard rename, hard delete; say it in the changelog.
- Delete before you abstract.
- Green gate before "done" — show the output.
- Stuck twice, stop and say so.
- No secrets, no personal data, no machine paths in committed files.
- Explain it so somebody who does not code can follow.
- KISS, YAGNI, DRY, SOLID, TDD, Clean Code, Clean Architecture — the criteria a review judges a diff by and a spec justifies a decision with, one line each.
- Monitoring, metrics and observability first. Nothing gets a URL until CI/CD, logs, traces, errors, health and data age, an external check, a second path and security all pass, and each one passes with a command rather than an assertion.
- A decision that always comes out the same is code, not a prompt. The third time the
same judgement resolves the same way it becomes a script, and the prompt that made it
goes away in the same commit. The script lives in this repository, has one check, runs
inside
just check, and fails closed. If it cannot fail closed, it stays a prompt and you write down why.
What this project is
A wheel on PyPI that carries twenty skills, four guards and a ten-verb CLI. One command places the skills and registers the guards in the settings file each surface already reads, so they are present in every project on a machine without a single file landing in any of them. It writes specs, plans, decisions and dated risk acceptances as plain text in the user's repository, and verifies them with a command that exits non-zero.
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.
- 3d ago First seen · 122 lines · 1,753 tokens per session scan A e71c26e7014a
ai-engineering AGENTS.md is an instructions file published in the GitHub repository arcasilesgroup/ai-engineering (54 stars, last pushed 4d ago), licensed Apache-2.0. It adds 1,753 tokens to every session, about $0.0088 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.
Other instructions, from other repositories
Inspectra copilot-instructions.md
Copilot instructions for Fascinax/Inspectra, covering inspectra — global copilot instructions, project overview, architecture, critical rules and output format.
Inspectra CLAUDE.md
Claude Code instructions for Fascinax/Inspectra, covering inspectra, quick reference, project structure, architecture rules and scoring.
Inspectra AGENTS.md
AGENTS.md instructions for Fascinax/Inspectra, covering agents, workflow overview — tier b (default), workflow overview — map-reduce (multi-agent), workflow overview — fusion (maximum recall) and audit modes.
code-security-skill CLAUDE.md
Claude Code instructions for Chiehyii/code-security-skill, covering code security skill — claude.md, what this skill does, how to use the search engine, full security report for any feature and get checklist for file upload.
apm architecture.instructions.md
Single canonical owner discipline: one authority per durable decision, guarded by a regression test + a static boundary check.
apm python.instructions.md
Python development guidelines.