.NET: Proposal: add an llms.txt

[BrethofAI]

Hi! We would like to offer Microsoft Agent Framework an llms.txt: a short, plain-markdown guide at the root of the repository that tells AI assistants what the project is and where its documentation lives (the llms.txt format). We drafted one with our open-source generator brethof-llms-txt; an AI model (GLM 5.3 Flash) wrote the summary and descriptions. It is below.

Would you like it as a pull request? If yes, reply here and we will open one. If not, close this issue and we will not ask again.

The proposed llms.txt (21 links)

# Microsoft Agent Framework

> An open-source, multi-language framework for building production-grade AI agents and multi-agent workflows in .NET, Python, and Go.

Supports Python and C#/.NET with consistent APIs, plus a separate Go SDK in microsoft/agent-framework-go. Provides orchestration patterns (sequential, concurrent, handoff, group collaboration), middleware, OpenTelemetry observability, YAML declarative agents, DevUI, and Foundry hosted agents. Install via Python or .NET packages; docs include quickstart, tutorials, user guide, and migration guides from Semantic Kernel and AutoGen.

## Docs

- [Microsoft Agent Framework README](https://github.com/raw/microsoft/agent-framework/main/README.md): Overview of the framework for building production-grade AI agents and multi-agent workflows in .NET and Python.
- [Welcome to the Agent Framework Community](https://github.com/raw/microsoft/agent-framework/main/COMMUNITY.md): Ways to get involved: GitHub discussions, issues, pull requests, and public community office hours schedules.
- [Responsible AI Transparency FAQs](https://github.com/raw/microsoft/agent-framework/main/TRANSPARENCY_FAQ.md): Answers about what the framework is, its capabilities, and intended uses.
- [Frequently Asked Questions](https://github.com/raw/microsoft/agent-framework/main/docs/FAQS.md): Steps to access nightly builds via GitHub Personal Access Token and NuGet configuration.

## Features

- [Durable Agents Have Moved](https://github.com/raw/microsoft/agent-framework/main/docs/features/durable-agents/README.md): Links to the new repository location for durable agent source, samples, and documentation.
- [Vector Stores and Embeddings](https://github.com/raw/microsoft/agent-framework/main/docs/features/vector-stores-and-embeddings/README.md): Design decisions for ported vector store and embedding abstractions from Semantic Kernel.
- [FIDES Implementation Summary](https://github.com/raw/microsoft/agent-framework/main/docs/features/FIDES_IMPLEMENTATION_SUMMARY.md): Deterministic prompt injection defense system using content labels, SecureAgentConfig, MCP auto-labeling, and data exfiltration prevention.
- [CodeAct .NET implementation](https://github.com/raw/microsoft/agent-framework/main/docs/features/code_act/dotnet-implementation.md): .NET CodeAct design with HyperlightCodeActProvider, provider-owned tool sets, swappable backends, and execution capability configuration.
- [CodeAct Python implementation](https://github.com/raw/microsoft/agent-framework/main/docs/features/code_act/python-implementation.md): Python CodeAct design with HyperlightCodeActProvider, provider-owned tool sets, swappable backends, and execution capability configuration.

## Optional

- [Contributing to Agent Framework](https://github.com/raw/microsoft/agent-framework/main/CONTRIBUTING.md): How to report issues, file pull requests, and rules for contributing new language implementations.
- [Architectural Decision Records (ADRs)](https://github.com/raw/microsoft/agent-framework/main/docs/decisions/README.md): What ADRs are and how to create, number, and review decision records using the templates.
- [Agent Run Responses Design](https://github.com/raw/microsoft/agent-framework/main/docs/decisions/0001-agent-run-response.md): Design of agent run response abstractions covering messages, tool activities, reasoning, handoffs, and streaming updates.
- [Agent Tools](https://github.com/raw/microsoft/agent-framework/main/docs/decisions/0002-agent-tools.md): Design decision on a unified tool abstraction, provider-specific tool options, error handling, and fallbacks for custom tools.
- [Agent OpenTelemetry Instrumentation](https://github.com/raw/microsoft/agent-framework/main/docs/decisions/0003-agent-opentelemetry-instrumentation.md): Decision on OpenTelemetry instrumentation for agents, covering semantic conventions, token usage, traces, and non-intrusive optional telemetry.
- [Azure.AI.Agents.Persistent` package Extensions Methods for Agent Framework](https://github.com/raw/microsoft/agent-framework/main/docs/decisions/0004-foundry-sdk-extensions.md): Decision on where extension methods for creating AIAgent from PersistentAgentsClient should live, comparing package placement options.
- [Python Package design for Agent Framework](https://github.com/raw/microsoft/agent-framework/main/docs/design/python-package-setup.md): Python package structure with tier 0, tier 1, and tier 2 components and flat imports from agent_framework.
- [Agent Framework / Foundry SDK Alignment](https://github.com/raw/microsoft/agent-framework/main/docs/specs/001-foundry-sdk-alignment.md): Specification clarifying positioning of Foundry SDK versus Agent Framework SDK, their goals, and combining both in orchestrations.
- [Python protocol helpers and optional execution state](https://github.com/raw/microsoft/agent-framework/main/docs/specs/002-python-hosting-channels.md): Python implementation plan for protocol helper functions and optional state holders in agent-framework-hosting, with goals and non-goals.
- [NET hosting: OpenAI Responses protocol helpers and optional execution state](https://github.com/raw/microsoft/agent-framework/main/docs/specs/003-dotnet-hosting-protocol-helpers.md): Helpers for exposing an AIAgent or workflow over the OpenAI Responses protocol in your own ASP.NET Core routes.
- [Feature-usage telemetry via an accumulating bitmask](https://github.com/raw/microsoft/agent-framework/main/docs/specs/004-feature-usage-telemetry.md): Design for feature-usage telemetry via an accumulating bitmask on the User-Agent, with allowlisted pipelines and opt-out.
- [Python function-calling loop contract and validation matrix](https://github.com/raw/microsoft/agent-framework/main/docs/specs/004-python-function-calling-loop.md): Required behavior and validation coverage for the Python function-calling loop, including approvals, streaming, errors, and serialization.

— BrethofAI

[BrethofAI]

One thing we should have said up front:

No maintenance for you. You would not have to keep it up to date. The free Brethof llms.txt app does it for public repositories: once a week it checks your docs and, if they changed, updates the file (a moved or deleted page is fixed the same day), in one of two ways, your choice:

• You approve each update: it opens a pull request, and nothing changes until you merge it.
• Fully automatic: tick one box and updates are committed directly, with nothing for you to do.

You can switch between them at any time. Without the app the file simply stays as it is.

— BrethofAI