eval-guide
microsoft/eval-guide/.github/copilot-instructions.md
This repository is an AI agent evaluation toolkit for Copilot Studio. It provides structured methodologies for planning, generating, running, and interpreting AI agent evaluations.
How real projects brief GitHub Copilot, from .github/copilot-instructions.md.
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
microsoft/eval-guide/.github/copilot-instructions.md
This repository is an AI agent evaluation toolkit for Copilot Studio. It provides structured methodologies for planning, generating, running, and interpreting AI agent evaluations.
microsoft/terraform-provider-fabric/.github/copilot-instructions.md
This repository is terraform-provider-fabric — the official Terraform provider for Microsoft Fabric.
microsoft/copilot-for-eclipse/.github/copilot-instructions.md
This is the GitHub Copilot for Eclipse plugin - an Eclipse IDE extension that brings AI-powered code completions, chat assistance, and intelligent coding features to Eclipse users. You MUST verify compilation and code quality before declaring work complete! When code changes are complete (after creating, editing, or refactoring code files), automatically verify that the changes follow project best practices: Action: After completing code changes, review against this checklist and fix any violations before declaring work complete. The project follows Eclipse…
microsoft/VirtualClient/.github/copilot-instructions.md
VirtualClient is a cross-platform benchmarking framework by Microsoft. It runs profile-driven performance benchmarks, stress tests, and qualification workloads on systems (particularly Azure VMs), collecting structured metrics. Supports Windows and Linux on x64 and ARM64. All workloads, dependencies, monitors inherit VirtualClientComponent. Constructor: (IServiceCollection, IDictionary<string, IConvertible>). Lifecycle: IsSupported → InitializeAsync → Validate → ExecuteAsync → CleanupAsync. External binaries run through IProcessProxy (wraps System.Diagnostics.Process). Output captured via ConcurrentBuffer. Tests use InMemoryProcess. Version is read from the VERSION file. Override with VCBuildVersion env var.
microsoft/micro-frontend-react/.github/copilot-instructions.md
Do not upgrade these beyond their UMD-available versions: - react / react-dom / react-is: 18.3.1 (last version with official UMD builds; React 19 dropped UMD) - react-router-dom: 5.3.4 (v6+ dropped UMD; v5 API: Switch, exact, render prop, withRouter) - styled-components: 5.3.11 (v6 dropped UMD)
microsoft/AntiSSRF/.github/copilot-instructions.md
You are a coding assistant AI that follows the company’s Security Development Lifecycle (SDL) guidelines for secure coding. Always apply the following practices when generating source code: - Use approved frameworks & libraries: Leverage well-vetted, company-approved languages, frameworks, and APIs for security functionality. Do not write custom crypto or auth logic if a standard solution exists. - Validate all inputs: Treat all inputs as untrusted. Implement strict input validation and sanitization (allow-list acceptable values or formats). Reject or sanitize data…
microsoft/vscode-copilotstudio/.github/copilot-instructions.md
Shared AI-agent guidance for this repository lives in AGENTS.md. GitHub Copilot users and agents should follow AGENTS.md.
microsoft/vscode-powerquery/.github/copilot-instructions.md
This is a multi-package monorepo (root, client/, server/, scripts/). Each has its own node_modules and package.json. The root npm install runs npm install-clean in all sub-packages via postinstall. Before declaring a task complete, always verify that changes pass npm run build, npm run lint, and npm run test:server with no errors. The PR gated build enforces all three. This is a VS Code extension providing Language Server Protocol (LSP) support for the Power Query / M formula language. Client (client/src/extension.ts):…
microsoft/haste/.github/copilot-instructions.md
Always run tests before marking a task as complete. Always run lint before committing changes. All feature work, refactors, and architecture decisions are driven by specs in spec/.
microsoft/Nonprofits/NonprofitDataSolutions/FabricWorkload/.github/copilot-instructions.md
This is a Microsoft Fabric Workload (WDK v2) for nonprofit organizations. It deploys a medallion architecture (Bronze → Silver → Gold) for fundraising analytics via a package installer UI inside Microsoft Fabric. @src/* → app/*, @clients/* → app/clients/*, @controller/* → app/controller/*, @components/* → app/components/*, @originalInstaller/* → app/items/PackageInstallerItem/*, @nds/* → app/items/NonprofitDataSolutions/* Read the relevant .ai/context/ files for detailed guidance: Every component must follow the pattern: ComponentName/ with index.ts, ComponentName.tsx, ComponentName.types.ts, ComponentName.styles.ts, and optionally ComponentName.model.ts. See .ai/component-structure-standards.md for full rules.
microsoft/PromptKit/.github/copilot-instructions.md
When a user asks you to perform a task that this library supports (writing requirements, investigating bugs, reviewing code, etc.), use the /promptkit skill (or read and execute bootstrap.md). It is the entry point that guides component selection, parameter gathering, and prompt assembly. Do not try to answer the task directly — assemble the right prompt first. PromptKit is a composable prompt library, not a software application. There is no build system, no runtime, and no application code. Every file…
microsoft/vscode-isort/.github/copilot-instructions.md
This is a VS Code extension that wraps isort for Python import sorting. It's a dual-language project: TypeScript for the VS Code client, Python for the LSP server. The extension operates in two modes controlled by the isort.serverEnabled setting: 1. Server mode (default): TypeScript client starts the Python LSP server as a child process. The server handles all sorting/diagnostics via LSP protocol. 2. Server-less mode: TypeScript client runs isort directly via script_runner.py, providing code actions and diagnostics without the LSP…
microsoft/kanagawa/.github/copilot-instructions.md
No summary in the file. Open it to read it.
microsoft/edge-ai/.github/copilot-instructions.md
Comprehensive coding guidelines and instructions for edge ai - Brought to you by microsoft/edge-ai
microsoft/teams.ts/.github/copilot-instructions.md
Follow the shared repository guidance in AGENTS.md.
microsoft/durabletask-mssql/.github/copilot-instructions.md
This repository implements the Microsoft SQL Server and Azure SQL storage provider for the Durable Task Framework and Durable Functions. It contains: The provider persists orchestration and entity state, coordinates distributed workers, and supports multiple independently deployed applications sharing one database. Correctness, rolling-upgrade compatibility, and predictable SQL performance take priority over convenience. Published table-valued parameters are wire contracts for independently deployed apps sharing a database schema. Never change or recreate a published TVP in a minor or patch release, even…
microsoft/Microsoft-AI-Decision-Framework/.github/copilot-instructions.md
Overall guidelines about the project.
microsoft/cat-agent-skills/.github/copilot-instructions.md
This repo is a gallery of reusable Agent Skills. Contributors add one submission under submissions/<slug>/ (a metadata.json sidecar plus exactly one payload — an unpacked SKILL.md skill, a .zip, or a Scout .json); CI generates the published page and download bundle. See CONTRIBUTING.md and submissions/README.md for the full rules. When reviewing a pull request, check the following in addition to CI: These checks are a merge gate, not a copy-edit. Skills are human-authored prose and code; treat the author's wording,…
microsoft/webui/.github/copilot-instructions.md
You are working on WebUI, a high-performance server-side rendering framework written in Rust that operates without JavaScript runtimes. It separates static and dynamic content at build time into a binary protocol that enables fast rendering in any host language. Read and internalize these instructions at the start of every session. They are non-negotiable. Before suggesting or applying any change, read these files - they are the ground truth: Before creating any commit, run: This executes, in order: license-headers → fmt…
microsoft/agentic-agile-template/.github/copilot-instructions.md
Stub: The full Copilot instructions have moved. See the path below. This stub remains at .github/copilot-instructions.md because GitHub auto-discovers it from that path. The adapter's quickstart.md explains how to copy the full content here when adopting this template in a downstream repo. This project uses a three-layer adapter architecture: a universal core, platform-specific adapters (platform-adapters/), and optional tool-specific files like this one. See AGENTS.md in the repository root — it is the universal entry point for all AI agents and…
Repository-wide instructions for GitHub Copilot, kept in .github/copilot-instructions.md.
How to build and test, the project's structure, and the conventions Copilot should follow in suggestions and reviews.
No. Copilot's coding agent also reads AGENTS.md, so many projects keep the shared rules there.
Open a file to see its discussion. Reports from people and their agents are coming.