agentleFS
MicrosoftKnown publishermicrosoft/fluentui-react-native/packages/agentic/desktop-driver-mcp/AGENTS.md

fluentui-react-native / desktop-driver-mcp

AGENTS.md1.4k starsChanged 5 months agoMITView on GitHub

This package is the Model Context Protocol adapter for the Desktop Driver agent operations. It is the instruction root for this package and every descendant. Read the README, then the narrowest entry in the reference index. The operations themselves, their schemas, and their failure reasons belong to @fluentui-react-native/desktop-driver (its agent operations reference); this package only maps them onto MCP. - Stay a thin client. Every tool calls the service's operation routes through the typed client; never add a second execution…

Trending

Quiet since March.about the whole repository

Dec 2025: 0 mentionsJan 2026: 0 mentionsFeb 2026: 1 mentionMar 2026: 1 mentionApr 2026: 0 mentionsMay 2026: 0 mentionsJun 2026: 0 mentionsJul 2026: 0 mentionsAug 2026: 0 mentionsSep 2026: 0 mentionsOct 2026: 0 mentionsNov 2026: 0 mentions
Dec 2025Jun 2026Nov 2026
Mentions a month of the microsoft/fluentui-react-native repository that our weekly crawl found. There is no adoption data for this file yet.

Who's talking about microsoft/fluentui-react-native

The biggest 5 of 37 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 37

Talked about on GitHub and Bluesky (37 times).

On GitHub · 36

31 more on GitHub

On Bluesky · 1

The file

We split this AGENTS.md into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read the README, then the narrowest entry in the reference index.

    line 8
  2. Very specific:

    Run yarn format, yarn lint, yarn build, and yarn test in this package.

    line 43

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

Desktop Driver MCP agent instructions

This package is the Model Context Protocol adapter for the Desktop Driver agent operations. It is the instruction root for this package and every descendant.

Start here

Read the README, then the narrowest entry in the reference index. The operations themselves, their schemas, and their failure reasons belong to @fluentui-react-native/desktop-driver (its agent operations reference); this package only maps them onto MCP.

Invariants

  • Stay a thin client. Every tool calls the service's operation routes through the typed client; never add a second execution path, a host, or state the service does not own. Tool input schemas come from the driver's registered operation schemas, so the service and the adapter cannot drift apart.
  • stdout carries MCP protocol only. Diagnostics go to stderr. Never log the grant, the descriptor contents, or application text.
  • Connect only through a private connection descriptor, verified by the installed native helper before the grant is sent. Never launch or build the service or a helper, and never accept a URL or grant from a command line, environment variable, or the model.
  • Serialize acquisition, use, close, and disposal per MCP connection. Retain cleanup responsibility after transport loss; reverify discovery and never replay a failed mutation. A verified replacement invalidates old references. Fence a lost acquisition response; disposal also cleans up late acquisitions.
  • Delegate a distinct driver-only child after verified discovery. Retain it on same-incarnation/descriptor-principal rediscovery; never share or resume another connection's session. Disposal awaits child revocation, keeping terminal-only cleanup proof for lost responses and failed drains; unauthorized is not cleanup.
  • Tie every artifact link to its session. Artifact ids restart in each session.
  • Progress ticks come from the adapter and never reach the service, so a waiting model does not keep an idle session alive. Cancellation aborts the HTTP call, which the service treats as the caller leaving.
  • Messages the model reads are fixed or bounded. Connection failures use fixed diagnostics on stderr too; never echo discovery exceptions or descriptor contents.

Validation

Run yarn format, yarn lint, yarn build, and yarn test in this package. Changes to tool schemas also need the tool-list size bound in src/server.test.ts to hold. A change to driver operations is validated in the driver package first. JavaScript tests use the fake host and are not native qualification; record a live run against a real application for native claims. Add a patch changeset for this published package when its behavior or shipped guidance changes.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

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.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.