AGENTS.md examples from real projects

vllm-projectPipes a download into a shellAGENTS.md

vllm

vllm-project/vllm/AGENTS.md

These instructions apply to all AI-assisted contributions to vllm-project/vllm. Breaching these guidelines can result in automatic banning. Before proposing a PR, run these checks: Do not open one-off PRs for tiny edits (single typo, isolated style change, one mutable default, etc.). Mechanical cleanups are acceptable only when bundled with substantive work. - Pure code-agent PRs are not allowed. A human submitter must understand and defend the change end-to-end. - The submitting human must review every changed line and run relevant…

vllm-projectAGENTS.md

vllm / rust

vllm-project/vllm/rust/AGENTS.md

This project implements an alternative, Rust-native frontend to the vLLM Engine. It is still experimental and actively evolving. The Python frontend is the behavioral reference for the OpenAI-compatible API; see Python frontend parity. Requests flow HTTP/gRPC DTO → chat or text request → engine request. Dependencies point downward: lower crates know nothing about the crates above them. Put a rule in the crate that owns its semantics, even if only one protocol triggers it today. The remaining crates are tooling…

vllm-projectAGENTS.md

vllm / tests

vllm-project/vllm/tests/AGENTS.md

- Test real objects on the lightest real path. Construct vLLM objects with their real constructors; do not stand them in with SimpleNamespace or MagicMock, set private attributes, or patch methods to reach a state. Mock only external services. Pick the cheapest setup that exercises the change: a layer or kernel test does not need vllm_runner, and single-rank parallel state is enough when parallelism is irrelevant. A fix to user-visible behavior needs at least one regression test through LLM (vllm_runner)…

vllm-projectAGENTS.md

vllm-omni

vllm-project/vllm-omni/AGENTS.md

Read CONTRIBUTING.md before contributing. Use the relevant repository skill: precheck-pr for author self-checks, or review-pr for independent reviews. Before code changes, issue triage or PR reviews, read relevant review and issue lessons and verify their evidence against the current revision.

vllm-projectAGENTS.md

semantic-router

vllm-project/semantic-router/AGENTS.md

vLLM Semantic Router is an Envoy ExtProc router for LLM inference. A public entrypoint resolves to an isolated recipe; its signals and projections feed a decision and algorithm, which select a backend and recipe-scoped plugins. - Read the nearest AGENTS.md only for directories you change. - Use tools/agent/docs/change-surfaces.md when a user-visible router contract crosses configuration, runtime, deployment, or documentation. - tools/agent/domains.yaml is the single changed-path registry for ownership, minimum checks, CI jobs, images, and E2E profiles. Use make harness-check after…

No other project uses it yet

vllm-projectAGENTS.md

aibrix

vllm-project/aibrix/AGENTS.md

AIBrix is a Kubernetes-native platform for scalable GenAI inference infrastructure. - Understand the problem and define success before making a non-trivial change. Ground decisions in existing implementations, tests, and verified evidence; do not infer behavior from filenames. - Make the smallest coherent change. Preserve unrelated work and existing conventions; avoid speculative abstractions and unrelated cleanup. - Treat public contracts and compatibility as design constraints. Keep API definitions declarative and policy in the layer that owns it. Preserve API, HTTP, CLI,…

vllm-projectAGENTS.md

vllm-ascend

vllm-project/vllm-ascend/AGENTS.md

This document provides instructions for contributors to the vLLM Ascend project. Please read and follow these guidelines to ensure code quality, maintainability, and consistency. Read linked documentation when it applies to the current task, using the relevant sections. Continue authorized local implementation and validation while preparing the evidence needed for PR review. Define environment variables in the centralized env_variables dictionary in vllm_ascend/envs.py. Requirements: Never: Hardcode environment variable names throughout the codebase. Reference them from the central module using from vllm_ascend…

No other project uses it yet

vllm-projectAGENTS.md

vllm-metal

vllm-project/vllm-metal/AGENTS.md

Follow the contributor guide for setup and general checks. For Omni additions, follow the model ownership and layout. Before opening a PR, search open PRs by topic. When working from an issue, read its comments and search for PRs referencing it: Do not open a duplicate PR. Explain any materially different approach in the PR description. Run checks relevant to the change. Report the commands, results, and any checks you could not run in the PR description. Extend related test…

vllm-projectAGENTS.md

guidellm

vllm-project/guidellm/AGENTS.md

This file provides agents and LLMs with guidance on development practices within the GuideLLM repository. NOTE TO AI: This file is human maintained and SHALL NOT be edited by agents or any LLM. NOTE TO HUMANS: This file should be kept brief as it is loaded into every AI context. IMPORTANT: tox should be used before attempting to run individual tools. IMPORTANT: For every command uv run must be prepended to the command. Use these markers to categorize and run…

No other project uses it yet

vllm-projectAGENTS.md

recipes

vllm-project/recipes/AGENTS.md

No summary in the file. Open it to read it.

vllm-projectAGENTS.md

speculators

vllm-project/speculators/AGENTS.md

Read CONTRIBUTING.md before contributing. Install the git hooks once per clone, after installing the [dev] extras: Fill in the PR template's Purpose and Tests sections (see PULL_REQUEST_TEMPLATE.md). A reviewer should get the point in under 30 seconds: Training, eval, and vLLM-launch scripts write reproducibility artifacts so runs can be recreated later. Always pass --provenance-dir when running scripts/launch_vllm.py — point it at the associated run's output/checkpoint directory (e.g. --provenance-dir <output_dir>) so the vLLM provenance is co-located with the training/eval it belongs…

vllm-projectAGENTS.md

agentic-api

vllm-project/agentic-api/AGENTS.md

Instructions for AI coding agents working on this repository. Read .agents/skills/agentic-api-engineering/SKILL.md before implementing, refactoring, fixing, or reviewing code. It defines repository-wide discovery, reuse of existing typed extension points, Rust design and Clippy expectations, provider documentation/parity evidence, and the recorder/test workflow. Apply its shared preparation once. For Rust changes, use the bundled Rust skill; use the focused review skills where relevant. This repository is Rust-first under the vllm-project GitHub organization. - Rust -- primary and active implementation language at the repo…

vllm-projectAGENTS.md

vllm-xpu-kernels

vllm-project/vllm-xpu-kernels/AGENTS.md

This file is the shared guidance for AI coding agents working in this repository. Keep tool-specific entry points, such as CLAUDE.md, thin and point them here so the project rules do not drift. vllm-xpu-kernels is a vLLM component that provides optimized custom kernels for Intel GPUs (XPU). Kernels are written in SYCL/DPC++ and use oneDNN and SYCL-TLA where appropriate. The package registers PyTorch custom ops that vLLM can dispatch on Intel GPU hardware. - Prefer current build metadata over stale…

vllm-projectAGENTS.md

vllm-gaudi

vllm-project/vllm-gaudi/AGENTS.md

This is vllm-gaudi — the vLLM Hardware Plugin for Intel® Gaudi® AI accelerators. It is a plugin package (not a fork) that integrates Intel Gaudi HPUs with the upstream vLLM inference engine via the vLLM plugin architecture. See the README — Getting Started for installation instructions (vLLM checkout, plugin install, Docker). The plugin follows vLLM's pluggable architecture. For Intel® Gaudi® hardware details (MME, TPC, profiling, etc.), see the Intel® Gaudi® Documentation. Key subsystems: