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/projectdxai/labrat/agents-mdgit clone --depth 1 https://github.com/ProjectDXAI/labratWhat 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 | $0.00661 | $0.00661 |
| Opus 5 | $0.00331 | $0.00331 |
| Sonnet 5 | $0.00132 | $0.00132 |
| Haiku 4.5 | $0.00066 | $0.00066 |
Grade A, and why
labrat 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- labrat CLAUDE.md — 92% identical, 13 lines differ
How it starts
The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.
labrat repo
This is the repository root, not a runnable lab directory.
Use this file when working on labrat itself from Codex.
If Codex is opened inside a nested lab that has its own AGENTS.md, follow the lab-local file for runtime operation. This root file governs repo maintenance, scaffold consistency, releases, and docs.
Source Of Truth
- Product/setup docs:
README.md,program.md,docs/getting-started.md,docs/runners.md,docs/MODEL_GUIDANCE.md - Scaffolded lab UX:
templates/AGENTS.md,templates/CLAUDE.md,templates/.agents/skills/,templates/.claude/commands/,templates/agent_prompts/ - Canonical example lab:
examples/nlp-sentiment/research_lab/ - Packaging/versioning:
pyproject.toml,labrat/__init__.py,CHANGELOG.md
Interface Rules
- Keep Codex and Claude Code first-class. If a lab gets one operator surface, add the matching one for the other interface.
- Generated labs should ship
AGENTS.md,.agents/skills/,CLAUDE.md,.claude/commands/, andagent_prompts/. - Keep
AGENTS.mdshort enough to stay durable. Put longer repeatable Codex workflows in.agents/skills/. - Do not require a hidden local setup for basic operation. The runnable contract should live in the repo.
- When operator guidance changes, keep templates, the canonical example lab, and the docs in sync.
- Prefer
labrat ...examples in docs when working from the repo root. Keep the copiedscripts/*.pyflow valid inside labs.
Frontier Model Rules
- Use
docs/MODEL_GUIDANCE.mdfor model and prompt-process updates. - GPT-5.5 is a Codex host selection, not a lab runtime constant. Do not hardcode API model IDs into
labratfor Codex usage. - Prefer explicit completion and verification contracts over broad encouragement.
- Reserve high reasoning effort for design, audit, frame break, profile authoring, and release work.
- Keep root
AGENTS.mdfocused on repo maintenance; lab runtime details belong in lab-localAGENTS.mdand.agents/skills/labrat-operator/SKILL.md.
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 · 45 lines · 661 tokens per session scan A e8b73b5c7480
labrat AGENTS.md is an instructions file published in the GitHub repository ProjectDXAI/labrat (238 stars, last pushed 25d ago), licensed MIT. It adds 661 tokens to every session, about $0.0033 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
Svatah CLAUDE.md
Instructions for bayeslabs-rsi/Svatah, covering svatah, architecture overview, key concepts, workspace & state and experiment lifecycle.
agentic-rd-skill AGENTS.md
AGENTS.md instructions for TheStreamCode/agentic-rd-skill, covering repository instructions, purpose and sources of truth, repository invariants, implementation style and documentation synchronization.
sprawling AGENTS.md
Instructions for 2youg1/sprawling, covering agents.md — how work is done in this repository, the loop, read before you write, one change, five steps and the rules a machine holds.
sous-chef AGENTS.md
Instructions for tomascupr/sous-chef, covering sous-chef, map and working agreements.
agiwo AGENTS.md
Instructions for xhwSkhizein/agiwo, covering agents.md, how to use this file, truth source order (adr 0048 / 0049), logging and config hot reload.
AI-System-Design-Consultant CLAUDE.md
Instructions for deepanshu2711/AI-System-Design-Consultant, covering claude.md, what this is, running it, load-bearing typos — do not "fix" these paths and agent node pattern.