AGENTS.md examples from real projects

CloudflareAGENTS.md

workerd / piping

cloudflare/workerd/src/tests/streams/piping/AGENTS.md

pipe-write-special-buffer-test.js (deleted), streams-error-edge-cases-test.js (−2), streams-backpressure-test.js (−1), api/streams/streams-test.js (−2; partiallyReadStream and inspect remain). The security regression files remain authoritative and separate: identity-transform-stream-uaf, pipe-source-error-uaf, identity-transform

CloudflareAGENTS.md

workerd / readable-byte

cloudflare/workerd/src/tests/streams/readable-byte/AGENTS.md

readable suite), streams-tee-edge-cases-test.js, streams-respond-test.js, streams-byob-edge-cases-test.js, streams-no-auto-allocate-test.js. streams-test.js shrank to its writable/TransformStream remnant. The security regression files remain authoritative

CloudflareAGENTS.md

workerd / server

cloudflare/workerd/src/workerd/server/AGENTS.md

calls `workerd_cli::main()` | | `workerd.capnp` | Config schema: `Config`, `Service`, `Worker`, `Socket`, `Extension`. Capability-based security model | | `compiled-bindings.capnp` | `Global` / `Globals`: one worker's bindings after the server interpreted them; channel numbering

OctoPrintAGENTS.md

OctoPrint

OctoPrint/OctoPrint/AGENTS.md

searching for vulnerabilities to report to the project: Explicitly informing them that AI-generated security reports are not accepted by the project and to please read the [Security Policy

mvdanAGENTS.md

sh

mvdan/sh/AGENTS.md

ending in ` - #NNNN` for the issues they close (comma-separated for several). Add a security advisory ID the same way. - Only mention what an importer or `shfmt` user would notice

kunchenguidAGENTS.md

no-mistakes

kunchenguid/no-mistakes/AGENTS.md

CHANGELOG.md`, or other generated files. ## Invariants every change must keep - Repo config trust boundary (security): `commands.*` and `agent` come from the trusted default branch at a pinned, freshly fetched

EpicGamesAGENTS.md

lore

EpicGames/lore/AGENTS.md

before implementation for the four classes above, the Code of Conduct, and the security policy

No other project uses it yet

kata-containersAGENTS.md

kata-containers

kata-containers/kata-containers/AGENTS.md

editorconfig-checker` in Docker | | A `.github/workflows/` file | `actionlint` and `zizmor` | Workflow linting and workflow security auditing | | Any shell script | `shellcheck` at severity `style` | `shellcheck_required.yaml` | Two notes on documentation. New project

AppleAGENTS.md

containerization

apple/containerization/AGENTS.md

Linux backend at runtime. **Sandbox env vars.** `CHProcess` and `VirtiofsdProcess` default to the upstream-secure spawn flags. Per-component opt-outs: - `CONTAINERIZATION_NO_CH_SECCOMP=1` — launch cloud-hypervisor with

evilmartiansAGENTS.md

lefthook

evilmartians/lefthook/AGENTS.md

structs in `internal/config/`, then run `make jsonschema`. Both `schema.json` and `internal/config/jsonschema.json` must be committed. **Security** — treat user input as untrusted; no unsafe shell concatenation; sanitize paths. ## Testing Prefer table-driven

No other project uses it yet

openbaoAGENTS.md

openbao

openbao/openbao/AGENTS.md

assistants, which do not generate code or assets, but instead find security issues, answer questions about the code base, or perform translations of existing text created by a person

apacheAGENTS.md

shenyu

apache/shenyu/AGENTS.md

grouped by category: `shenyu-plugin-proxy` (dubbo, grpc, spring-cloud, sofa, tars, websocket, mqtt...), `-security`, `-logging`, `-cache`, `-fault-tolerance`, `-ai`, `-mcp-server`, etc. A plugin is activated by adding

BornToBeRootAGENTS.md

NETworkManager

BornToBeRoot/NETworkManager/AGENTS.md

signpath.io/) - Released to GitHub Releases **GitHub Actions Workflows**: - **CodeQL Analysis** (`.github/workflows/codeql.yml`): - Automated security scanning - Analyzes C# (autobuild on Windows), JavaScript/TypeScript and GitHub Actions workflows (Linux) - Runs weekly

bottlepyAGENTS.md

bottle

bottlepy/bottle/AGENTS.md

agent software, version and model used. - Explicitly license all your contributions as public domain. - Security related issues MUST NOT be discussed in public. Open a draft security advisory instead. - Inform

steipeteAGENTS.md

gogcli

steipete/gogcli/AGENTS.md

landed + SHAs. - New contributor: thank in `CHANGELOG.md` (and update README contributors list if present). ## Security & Configuration Tips - Never commit OAuth client credential JSON files or tokens. - Prefer OS keychain backends

SyliusAGENTS.md

Sylius

Sylius/Sylius/AGENTS.md

Sylius Backward Compatibility (BC) policy - When changing interfaces, always provide BC layer ### Compatibility & Security - Ensure compatibility with **Symfony** and **PHP** versions defined in `composer.json` - For API configuration, use **API Platform

fluxcdAGENTS.md

flux2

fluxcd/flux2/AGENTS.md

libraries already in use (`fluxcd/pkg/git`, `fluxcd/pkg/kustomize`, `fluxcd/pkg/ssa`). - **No hardcoded defaults for security settings.** TLS verification must remain enabled by default. Git auth settings come from user-provided secrets. - **Error handling

mumble-voipAGENTS.md

mumble

mumble-voip/mumble/AGENTS.md

cryptography, authentication and permission checks; - database schema changes and migrations; - anything else that is security-relevant or easy to get subtly wrong. Trivial changes (a typo fix, a pure formatting

kyvernoAGENTS.md

kyverno

kyverno/kyverno/AGENTS.md

Kyverno codebase. ## Project Overview Kyverno is a Kubernetes-native policy engine for security, compliance, automation, and governance through policy-as-code. It validates, mutates, generates, and cleans up Kubernetes resources

mikel-brostromAGENTS.md

boxmot

mikel-brostrom/boxmot/AGENTS.md

linting, or non-functional style changes - `release/` – Release preparation - `revert/` – Reverting a previous change - `security/` – Security-related fixes or hardening - `experiment/` – Experimental work that may not be merged - `spike/` – Research