AIBAST Agents Library is the stable Microsoft downstream for the RAPP stack (see CONSTITUTION.md for architectural principles): The repository also owns agents/@aibast-agents-library/, registry.json, buildregistry.py, rappai/, the production guide, and Microsoft governance files. These are not upstream Grail mirrors. Read beta/GOLDEN_PATH.md before changing the beta. The beta is a chat-driven educational tool: users learn AI by watching GitHub Copilot build, hotload, run, test, explain, screenshot, and record real RAPP capabilities. External AIs can visibly drive the same Brainstem or Brain Surgeon…
This project uses the Spec-Kit agentic development process. Every change MUST follow these rules. Violations will be caught by CI and rejected. This repo uses spec-kit for structured feature development: When assigned an issue: When implementing a task autonomously: 1. Find the relevant spec — Check specs/ for the feature spec matching your issue. Read spec.md, plan.md, tasks.md, and data-model.md before coding. 2. Read the constitution — specs/constitution.md (v1.7.0) defines 15 mandatory architectural principles. Key constraints: - I. API-First: All…
This VS Code extension manages AI assistant catalog resources (chat modes, instructions, prompts, tasks, MCP servers) across multiple repositories. It provides centralized discovery, activation, and "Hats" (preset) functionality for AI resources that enhance GitHub Copilot workflows.
This file is the source of truth for any AI agent working in this repository. Address the human as Mr. Das (an alt of Iron Man), "Sir", or something similar. crap4csharp is a high-fidelity C# port of the Java tool crap4java (read-only sibling at ../crap4java) — a CRAP-metric analyzer for C# projects. Project design (intent, locked choices, the CRAP formula) lives in docs/decisions.md; per-feature plans in docs/features/<feature>.md. 0. All agents: - Crisp, high-signal communication. No verbosity; don't repeat the human's…
This is a multi-sample repository. Each sample folder at the repo root is a completely independent project with its own language, dependencies, configuration, and deployment target.
These instructions apply to any AI agent (Copilot CLI, Copilot Chat, Claude, Cursor, etc.) contributing to this repository. Start from AGENTS.md for the full router of context, conventions, and the build/test loop. - This is a TypeScript GitHub Action (src/index.ts → bundled to dist/ via ncc). - Code style, structure, and test patterns: docs/conventions.md. - GitHub Actions conventions (action.yml, inputs/outputs, secrets, workflow security): docs/github-actions.md. - Test framework and @actions/* mocking pattern: docs/testing-patterns.md. - Build/test loop: npm run build, npm run…
python-lsp-compare is a benchmark and regression harness for Python LSP servers. It downloads servers, runs LSP scenarios over stdio JSON-RPC, captures timing/payload metrics, and produces comparison reports. Pure Python, zero external dependencies. All source lives under src/pythonlspcompare/. Key modules and data flow: - **server_download.py* — Downloads servers from GitHub releases (ServerSpec) or installs into isolated venvs from PyPI (PypiServerSpec). Caches binaries under .python-lsp-compare/servers/. Version checks cached 24h in versions.json. Functions return tuple[Path, str] (executable path + version label). - *server_configs.py**…
This file provides guidelines for GitHub Copilot and other AI assistants when helping with this repository. All tools must output data in JSON format with standardized fields for consistency and interoperability. Every parser tool must use this exact structure for all output records: Every tool output must include these four required fields: All parser-specific data must be contained within the message field as a JSON object. This provides: - Consistent top-level structure across all parsers - Clear separation between metadata…
Versioning is dynamic via setuptools-scm from git tags. This is a keyring backend plugin (~435 lines) that authenticates pip/twine against Azure DevOps Artifacts feeds using pure Python — no .NET, no MSAL. Authentication flow (_backend.py → _provider.py → sessiontoken.py): Provider chain (provider.py): - Providers implement the TokenProvider Protocol (structural subtyping — no base class inheritance) - runchain() tries each provider in order, catches all exceptions, and returns the first successful token - Default chain: ["azurecli", "azureidentity"] — overridable via ARTIFACTSKEYRINGNOFUSS_PROVIDER…
Trust these instructions first. Only search the repo if information here is incomplete or found to be incorrect. Kompli is a Linux security compliance engine derived from Azure OSConfig. The repository keeps the ComplianceEngine module and the direct dependencies needed to build, test, package, and fuzz that module. The code is written in C11/C++11, uses CMake and vcpkg, and targets Linux only. Build outputs include build/adapters/mc/complianceengine/libOsConfigResourceComplianceEngine.so and build/modules/bin/complianceengine.so. Tests are registered from the retained common libraries and src/modules/complianceengine/tests/. Some tests…
These instructions guide AI coding agents (and the developers driving them) when working on this codebase. They are not the end-user KQL assistant prompt — that lives in src/Client/instructions.md and is shipped to users at runtime. Keep the two separate. For the full "why" behind everything below, read ARCHITECTURE.md. When a change contradicts these rules, prefer the architecture doc and update both. A VS Code extension that mimics the desktop Kusto Explorer app: edit multi-query .kql files, run queries, and…
Whenever making any Azure MCP server calls, you should always use the subscription ID of 5669097f-9a99-4354-b528-addefc61d776 and tenant ID of 16b3c013-d300-468d-ac64-7eda0820b6d3. Always assume Azure AD credentials when trying to connect to Azure services via MCP.
This file is the root of the agent governance framework. Individual agents have their specific governance files. These instructions and guardrails are sacrosanct. Never bypass or override them. Do not overdo things. 0. General principles: - When writing English (docs, code comments, Markdown, messages): - Use simple, precise language. Do not try to sound smart. - Use the fewest words that preserve meaning and accuracy. - Do not repeat the human's words. - For Markdown, follow .github/skills/markdown.md. - Do not…
This repository is a read-only-by-default Azure capacity/enablement toolkit. The full agent guidance lives in ../AGENTS.md — read it before running anything. Non-negotiable rules: - READ-ONLY BY DEFAULT. With the analysis scripts, never create, modify, delete, scale or restart any Azure resource — the only writes are local files under output/. No az ... create/update/delete/set, no New-Az* / Set-Az* / Remove-Az, no terraform apply, no kubectl apply/delete on your own initiative. - *One opt-in write tool.* Deploy-QuotaGroups.ps1 (with New-QuotaGroupConfig.ps1) provisions quota…
Before writing any Azure / Microsoft Foundry SDK code, or before describing a fast-moving Azure / Foundry capability: 1. Search Microsoft Learn through the microsoft-docs MCP server (the MS Learn MCP server) for the current API surface, SDK syntax, and product guidance. Use foundry-mcp as well for Foundry-native operations such as model catalog, agents, toolboxes, knowledge bases, and evals. Foundry features here are fast-moving and many are preview — never rely on memorized signatures. 2. Load the matching skill from…
MicrosoftCopilot instructions