agentleFS
Sign inSign up

AGENTS.md examples from real projects

How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.

Most stars first · from page 25Worked for most · soon
MicrosoftAGENTS.md

amplifier-app-simulated-user-research

microsoft/amplifier-app-simulated-user-research/AGENTS.md

Automated product-audit rounds: seed a scratch instance, drive real-browser persona sessions + design reviews, synthesize an evidence-tiered findings spec behind a human gate. Read PRINCIPLES.md before designing changes; SMOKE_TESTS.md before verifying. - The CLI command is amplifier-simulated-user-research — never abbreviate it. The retired four-letter acronym of this tool's name (a‑s‑u‑r) is banned everywhere: code, docs, messages, commits, PRs. - The user-facing browser-node bundle name is simulated-user-research-browser-node. Internal .dot param names (surrepodir) are grandfathered — leave them. pipelines/simulated-user-research.dot is the sole…

12mo agoDiscuss
MicrosoftAGENTS.md

amplifier-bundle-amplifier-tester

microsoft/amplifier-bundle-amplifier-tester/AGENTS.md

@README.md @bundle.md @agents/setup-digital-twin.md @agents/validator.md @context/amplifier-ecosystem-map.md

124d agoDiscuss
MicrosoftAGENTS.md

amplifier-bundle-memory

microsoft/amplifier-bundle-memory/AGENTS.md

Vision-first, contract-driven (Converge). docs/VISION.md governs the contracts; the contracts govern the code. A change that contradicts a contract lands in the contract first (proposal → owner's word), then in code. 1. Simplicity is the requirement, not a style. Before adding a mechanism, name the cost that was actually paid without it. The predecessor shipped a resident server, an evidence ladder, tombstones, fingerprints, semantic clustering, a pending pool, ceilings, and a 63-row ledger — and an empty store. The bar for…

117d agoDiscuss
MicrosoftAGENTS.md

amplifier-bundle-redaction

microsoft/amplifier-bundle-redaction/AGENTS.md

This repo is a bundle with three layers: a zero-dependency library at the root, a thin hook adapter under modules/, and the behavior that wires them together. Keep these layers separate — that separation is the entire point of this repo's shape. Do not collapse these back into a nested modules/redaction/ layout. The library moved to the repo root deliberately (see amplifier-module-hooks-redaction for the "before" shape this replaced) so that a consumer can import redaction out-of-process without going through Amplifier…

122d agoDiscuss
MicrosoftAGENTS.md

amplifier-module-loop-live

microsoft/amplifier-module-loop-live/AGENTS.md

The core package must not import Amplifier CLI, Amplifier Converge, provider SDKs, or host adapters. Optional application integration belongs under adapters/. Preserve ordinary finite execution when the host supplies no live runtime. Preserve input/call/session identities, approval boundaries, and truthful cancellation. Interrupted work is never automatically replayed. Use Amplifier Converge whenever naming that application. Keep credentials, transcripts, user settings, and private validation artifacts outside Git. Record tested revisions and distinguish fixture evidence from live model tests. Do not change provider protocol…

117d agoDiscuss
MicrosoftAGENTS.md

fsi-agentic-wealth-transfer

microsoft/fsi-agentic-wealth-transfer/AGENTS.md

Read .github/github-instructions.md and Software-Requirements-Document.md before changing the agent layer. Treat the SRD open questions as unresolved until a stakeholder answer is recorded; never replace a # TODO [Qn] with an assumption. The active Milestone 1 replacement is deployed as ai-transfer-agent version 8 in East US 2 under fsi-wt-dev; prior North Central US version 6 is retained as historical deployment evidence. Its immutable audit system of record is WORM Blob only. Do not add SQL, Fabric, Cosmos DB, live ACATS submission,…

12mo agoDiscuss
MicrosoftAGENTS.md

acs-resource-explorer

microsoft/acs-resource-explorer/AGENTS.md

Purpose: Automate the assessment of retiring Azure Communication Services (ACS) resources — scanning subscriptions, detecting active usage across all retiring channels, analyzing migration urgency, and generating reports with migration guidance.

09d agoDiscuss
MicrosoftAGENTS.md

amplifier-app-tui

microsoft/amplifier-app-tui/AGENTS.md

Read docs/VISION.md, contracts/, notes/CONVERGE.md, notes/PLAN.md, notes/ENGINE-BOUNDARY.md and SMOKETESTS.md before changing behavior. Keep runtime imports in host/composition and their app-policy adapters; widgets consume identified events only. Independent modules may import kernel…

010d agoDiscuss
MicrosoftAGENTS.md

amplifier-bundle-converge

microsoft/amplifier-bundle-converge/AGENTS.md

Read these before your first edit. They apply to every session working here — human, Amplifier, or another tool's coding agent — and they are the same rules the people here follow. Nothing on this page needs the Amplifier bundle. Read next: PINS.md (hard facts), docs/CONTRACTS-README.md (what a contract is and how to change one). New to this method, or taking it to another repository? docs/ADOPTING.md is the first day start to finish — install, check that it took, start…

08d agoDiscuss
MicrosoftAGENTS.md

amplifier-bundle-imagegen

microsoft/amplifier-bundle-imagegen/AGENTS.md

Own the portable image capability, its module, skill, context and composition. Keep application setup, UI/artifact presentation and live account qualification in the consuming host. Runtime code must not depend on…

05d agoDiscuss
MicrosoftAGENTS.md

amplifier-bundle-work

microsoft/amplifier-bundle-work/AGENTS.md

This repository owns portable Work composition, operating instructions, and composition tests. Keep runtime implementations in their module repositories. Application-specific setup, browser acceptance, and preview launchers belong in the consuming application. Do not add a dependency on an application here. Preserve provider neutrality, explicit delegation policy, and branch-tracking module sources. Do not pin bundle or module sources to commits or tags. Record tested revisions in validation receipts, without turning them into source constraints. Keep credentials, user settings, transcripts, and live test…

011d agoDiscuss
MicrosoftAGENTS.md

amplifier-skill-revisioner

microsoft/amplifier-skill-revisioner/AGENTS.md

VISION.md is the entry point for what this project is converging toward. The skills under skills/ are the implementation. For local development, use your agent's tools to load or read skills from skills/. The project has one word for the thing it works on: a risky assumption. Use it, and the terms below, in skill text, script output, UI copy, and when reporting to the user. Introducing a synonym costs the reader a translation step every time they meet it,…

021d agoDiscuss
MicrosoftAGENTS.md

amplifier-smart-tool-digital-twin-universe

microsoft/amplifier-smart-tool-digital-twin-universe/AGENTS.md

docs/00-vision.md is the highest-authority document in this repository. When a change conflicts with it, the vision wins; raise the conflict rather than silently overriding it. Every capability lives in amplifierdigitaltwinuniverse. No capability may exist only in the CLI (cli.py, catalog.py). Add a new capability to the library first, export it from _init__.py, then wire the CLI to it. The DTU profile schema, its loader, and its launch semantics are owned by amplifierbundledigitaltwinuniverse. profiles.py and environments.py delegate to it. Never reimplement…

030d agoDiscuss
MicrosoftAGENTS.md

bobpgsqldemos / caldova-logistics-agent

microsoft/bobpgsqldemos/caldovalogistics/demo4-foundry-agent/caldova-logistics-agent/AGENTS.md

This project is a Microsoft Foundry hosted agent — a containerized AI agent that runs in Foundry Agent Service. The platform handles containerization, hosting, security, scaling, and observability so you can focus on agent logic. The Azure Developer CLI (azd) manages the full lifecycle: If you are in VS Code, read the vscode-microsoft-foundry skill first. Install the Microsoft Foundry Skill for guided deployment, evaluation, and troubleshooting workflows. Direct install (preferred, works with any coding agent): Or install the Azure Skills…

015d agoDiscuss
MicrosoftAGENTS.md

brewlet / integration-tests

microsoft/brewlet/integration-tests/AGENTS.md

The strict scenarios in e2e/live/ are independent of the legacy tiers below. Run python3 integration-tests/e2e/live/hpa.py or python3 integration-tests/e2e/live/admission.py from the repository root. They create their own uniquely named clusters and registries; never pass a shared kube context or run the tier reset helper for them. Mandatory assertions fail instead of skipping. .github/workflows/e2e.yml has separate scheduled/manual live jobs, each running twice on fresh clusters; it has no push/PR triggers. See the live-validation runbook for release pins, capacity, load leases, stabilization, fixture-only…

012d agoDiscuss
MicrosoftAGENTS.md

frontier-ai-governance-rvas

microsoft/frontier-ai-governance-rvas/AGENTS.md

Remember the law of diminishing returns to decide if something should fit or not here. Before drafting, editing, or returning repository prose, invoke /humanize-writing. This includes technical documentation, session kits, runbooks, Markdown, slide content, and user-facing explanations. In this repository, use the skill to simplify wording and remove AI-patterns from technical content. Its general technical-writing exclusion does not apply here. Use the clear-thinker voice unless the user requests another voice. Before finishing, complete both checks: 1. Apply the prose simplification…

027d agoDiscuss
MicrosoftAGENTS.md

frontier-ghcp-ttt-rvas

microsoft/frontier-ghcp-ttt-rvas/AGENTS.md

Before drafting, editing, or returning repository prose, invoke /humanize-writing. This includes technical documentation, session kits, runbooks, Markdown, slide content, and user-facing explanations. In this repository, use the skill to simplify wording and remove AI-patterns from technical content. Its general technical-writing exclusion does not apply here. Use the clear-thinker voice unless the user requests another voice. Before finishing, complete both checks: 1. Apply the prose simplification rules in this file. 2. Apply the /humanize-writing AI-pattern dictionary and revise the draft until…

022d agoDiscuss
MicrosoftAGENTS.md

opentelemetry-distro-browser

microsoft/opentelemetry-distro-browser/AGENTS.md

Avoid creating new Markdown (.md) files as part of routine work. Prefer updating existing documentation, subject to the README restrictions below. Only add a new document when explicitly requested by the user or when it adds clear, lasting value that existing documentation cannot cover. Do not create summary, planning, progress, or implementation-note Markdown files unless requested. Do not add README.md files inside source folders. This repository has one root README.md. Document code in TSDoc on the declarations themselves. Do not…

08d agoDiscuss
MicrosoftAGENTS.md

verbose-disco

microsoft/verbose-disco/AGENTS.md

Primary instructions for AI agents operating in this repository.

07mo agoDiscuss
MicrosoftAGENTS.md

XBOX-GDK-dotnet

microsoft/XBOX-GDK-dotnet/AGENTS.md

The dotnet implementation of the Microsoft GDK idiomatic projection. It is one of eight sibling language repositories split out of the gdk-projections-plans meta repo. The authoritative specification for this repository is plans/dotnet.md in the gdk-projections-plans meta repo, alongside the nine shared, language-agnostic reference docs under reference/ (gdk-surface.md, xuser-pilot.md, state-change.md, multiplayer-pilot.md, roadmap.md, security-privacy.md, compliance.md, glossary.md, testing.md). The plan follows a shared 15-section template; §4 (the mapping table) is the heart, §9 is the acceptance slice (the XUser pilot, plus the PFMP…

07d agoDiscuss
CLAUDE.md vs AGENTS.md

About AGENTS.md

What is AGENTS.md?

An open format for instructions to coding agents, read by Codex, Cursor and others. Think of it as a README written for agents.

Where does it go?

At the repository root, with more specific files in subdirectories. Agents read the one closest to the file they're editing.

What should it contain?

Setup and test commands, code style, and the rules a new contributor would need to know.

Does Claude Code read it?

Claude Code reads CLAUDE.md. A one-line CLAUDE.md that points at AGENTS.md covers both.