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/atlas-php/atlas/agents-mdgit clone --depth 1 https://github.com/atlas-php/atlasWhat 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 | $0.04577 | $0.04577 |
| Opus 5 | $0.02289 | $0.02289 |
| Sonnet 5 | $0.00915 | $0.00915 |
| Haiku 4.5 | $0.00458 | $0.00458 |
Grade A, and why
atlas 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.
How it starts
The opening of the file, as written. The whole thing — 391 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS
Rules for every agent working in this repository. These rules are law; where they conflict with your general habits, this file wins.
This is an open-source Laravel package — a unified AI SDK (Atlas v3) that owns its own provider layer.
Everything here ships into strangers' applications: the public API is a contract, every dependency is an
imposition on every consumer, and the documentation is the product's front door. The what & why lives in
.knowledge/BRIEF.md; the knowledge map is .knowledge/README.md. This file defines how you build here.
Before you work
Load light; pull depth only when the task needs it.
- Always read first:
.knowledge/BRIEF.md(what & why),.knowledge/CODEMAP.md(where things are),.knowledge/MEMORY.md(current friction)..knowledge/README.mdmaps the rest. - On demand, when the task enters an area:
.knowledge/prd/(ratified contracts — source of truth),.knowledge/prd-drafts/(proposals),.knowledge/research/+.knowledge/references/(prior art, visual targets),.knowledge/guides/(how to write each doc + project how-tos). - How work flows:
research/->prd-drafts/->prd/; aprd/contract never cites a draft. New guaranteed behavior is aprd/row backed by a test — cite itsR-<AREA>-<n>in the code. Follow a doc's guide before writing or modifying it, and keep docs true in the same task. Runpython3 .knowledge/scripts/doc-lint .knowledgebefore finishing; scratch ->.knowledge/tmp/. - Read every file before editing it; search before writing new logic — reuse, extend, refactor.
- When the user raises a concern, investigate before contradicting — evidence, not a hunch.
- Read the relevant VitePress documentation before working on any module. Consumer docs are the source of truth — they override all assumptions.
Hard gates — require explicit approval
- Persisted state. Any change to schema, stored data, or migrations is confirmed first.
- Dependencies. Do not add, remove, or major-version-bump a package without approval.
- Deletions. Do not delete files outside the task's immediate scope without approval.
- Commits. Do not commit or push unless told to.
- This file. Never modify
AGENTS.mdwithout approval; when approved, follow.knowledge/guides/docs-agents.md. If a rule seems wrong or missing, raise it.
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.
- 2d ago First seen · 391 lines · 4,577 tokens per session scan A 5471f66def47
atlas AGENTS.md is an instructions file published in the GitHub repository atlas-php/atlas (54 stars, last pushed 1mo ago), licensed MIT. It adds 4,577 tokens to every session, about $0.0229 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
best-of-Agent-Harnesses CLAUDE.md
Claude Code instructions for RyanAlberts/best-of-Agent-Harnesses, covering repo notes for ai coding agents, commit identity (required), branching and pushing (required), verifying contribution credit (required) and how the list is generated.
hindsight CLAUDE.md
Instructions for vectorize-io/hindsight, covering claude.md, project overview, development commands, local development (api + ui) and start both api server and control plane ui.
codedb AGENTS.md
AGENTS.md instructions for justrach/codedb, covering codedb agent guidelines, what codedb is (and isn't), review guidelines, pre-merge verification and security-sensitive areas.
claude-code-settings copilot-instructions.md
Instructions for feiskyer/claude-code-settings, covering claude.md, environment setup, required dependencies, configuration and skills.
agentor copilot-instructions.md
Instructions for CelestoAI/agentor: Read AGENTS.md in the repository root. It is the single source of contributor guidance for this project — architecture, development setup, code style, testing, and CI — and applies to Copilot exactly as written.
pi CLAUDE.md
Instructions for share-skills/pi, covering 强制, 内部目标（不可写入公开文件）, 跑分规范 and eval 驱动优化规则.