Oumi is a toolkit for building foundation models, including preparing data, fine-tuning models, evaluating them, and deploying inference endpoints. Researchers and developers use it with open-weight language and vision-language models such as Qwen and Gemma. The catalogue instructions support workflows around operating Oumi.
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/oumi-ai/oumi/agents-mdgit clone --depth 1 https://github.com/oumi-ai/oumiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/oumi-ai/oumi/agents-md)<a href="https://agentmods.dev/instructions/oumi-ai/oumi/agents-md"><img src="https://agentmods.dev/badge/instructions/oumi-ai/oumi/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00436 | $0.00436 |
| Opus 5 | $0.00218 | $0.00218 |
| Sonnet 5 | $0.00087 | $0.00087 |
| Haiku 4.5 | $0.00044 | $0.00044 |
Grade A, and why
oumi 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 5d 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 — 25 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent instructions
For coding agents (Claude Code, Codex, Cursor, and anything similar) working in this repository. Humans should read CONTRIBUTING.md — everything there applies to agents too.
Publication policy
Never publish without the developer's explicit, per-instance approval. Publishing means: pushing a branch (including new commits to an existing pull request branch), creating a fork or a remote repository, opening a pull request (draft included), or rewriting a published pull request's title or body.
This repository is public. Commit messages, branch names, pull request text, and code comments are visible to everyone the moment they leave the machine, and deleting them later does not un-publish them.
A general instruction to "work on this" is not approval to publish. Neither is a plan the developer approved, nor a task list they accepted. Approval is per publication, and it comes after they have seen what would go out.
Before publishing:
- Show the developer what would become visible — every commit message, the pull request title and body, and the lines the diff adds. Not a summary of them.
- Check that content for anything that should not be public: credentials and keys, internal hostnames and infrastructure identifiers, customer or partner names, people's contact details, and unreleased or business-internal information. Pattern matching alone misses names, so read the content as well as scanning it.
- Wait for the developer to say yes, having seen the above. If something was flagged, it comes out or they explicitly accept it first.
- Then publish, and share the resulting link.
If no developer is available to approve, keep the work on a local branch and say what is waiting. Do not publish and report it afterwards.
Working with contributions
Follow CONTRIBUTING.md for branch naming, the fork-and-pull-request flow, tests, and style. Run pre-commit before proposing a change. Every pull request still needs a review from a member of oumi-ai/oumi-staff.
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.
- 5d ago First seen · 25 lines · 436 tokens per session scan A 3025afe5b064
oumi AGENTS.md is an instructions file published in the GitHub repository oumi-ai/oumi (9,383 stars, last pushed yesterday), licensed Apache-2.0. It adds 436 tokens to every session, about $0.0022 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
ollama AGENTS.md
AGENTS.md instructions for ollama/ollama, covering agents.md and building.
ollama CLAUDE.md
Claude Code instructions for ollama/ollama: See AGENTS.md for the shared agent instructions for this repository.
unsloth-buddy AGENTS.md
AGENTS.md instructions for TYH-labs/unsloth-buddy, covering unsloth-buddy — agent definition, role, activation, how to run and lifecycle (7 phases + demo).
geval copilot-instructions.md
Instructions for geval-labs/geval, covering geval - ai coding instructions, project overview, repository structure, build and test and cli and exit codes.
graphiti AGENTS.md
AGENTS.md instructions for getzep/graphiti, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
autocontext CLAUDE.md
Claude Code instructions for greyhaven-ai/autocontext, covering claude.md, project overview, repository layout, commands and setup.