agentleFS
MicrosoftKnown publishermicrosoft/amplifier-smart-tool-reality-check/skills/reality-check/SKILL.md

reality-check

Skill2 starsChanged 9 days agoMITView on GitHub

Checks that an app works the way a real user would experience it and returns evidence for every finding. Drive it from the command line as `reality-check`, or from Python through `reality_check.lib`. Use whenever a fix or capability needs to be validated: once you have fixed a bug, or built or changed a feature, run a check of what changed before you report the work as done, so the report says what a user can actually do, not only what the code or tests suggest. Also use when asked to check, verify, or make sure an app works, and when the user mentions "reality check" directly.

14 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 1 is distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 12 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    Read each capability's own skill with reality-check <command> --help before calling it: it carries the arguments, a worked invocation, the …

    line 38
  2. Very specific:

    If open_app is not among your tools, the MCP server is not registered or the host has not reloaded it: reality-check mcp --help says how to …

    line 46
  3. Very specific:

    Keep the tool up to date for the user by default if the permission mode allows it.

    line 53
  4. Very specific:

    If it is newer than reality-check --version, run the install command again with that tag and the extras uv tool list --show-extras reports.

    line 61
  5. Very specific:

    Update the skill with npx skills add microsoft/amplifier-smart-tool-reality-check#<tag>, adding -g for global installs.

    line 63
  6. Distinctive:

    Do not try to get around restrictions or operate against the user's preferences to update tools (or skills).

    line 64

“Common” is measured: kept by at least 6 in 10 of the 12 edited versions of this file we could read. “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
SKILL.md

Using reality-check

Checks that an app works the way a real user would experience it, on your machine, in a Digital Twin Universe, or in one it sets up for you, and returns evidence for every finding.

Install

# as a CLI, choosing the agent provider(s) you want
uv tool install "reality-check[all | copilot | amplifier-agent | codex | claude] @ git+https://github.com/microsoft/amplifier-smart-tool-reality-check@v0.8.1"

# as a library, from another project
uv add "reality-check[all] @ git+https://github.com/microsoft/amplifier-smart-tool-reality-check@v0.8.1"

# once, without installing
uvx --from "reality-check[all] @ git+https://github.com/microsoft/amplifier-smart-tool-reality-check@v0.8.1" reality-check --help

# once, to give checks everything they can use
reality-check install

# once, to show checks inline: register `reality-check mcp` as a stdio MCP server in the
# client, then reload its MCP servers. `reality-check mcp --help` has the details.

Use it

Run reality-check --help. It prints the tool's skill: when to use it, every capability, which files to read, etc. Read each capability's own skill with reality-check <command> --help before calling it: it carries the arguments, a worked invocation, the result, etc. The main smart capability is reality-check check-software --request "Check that a new user can sign up and send a message", which uses an Agent SDK like Copilot SDK or Amplifier Agent to validate software. Checks take at least a few minutes, with more complex or generic asks possibly taking much longer. reality-check check-software --help has the details.

To show the user their checks, call the MCP tool open_app with the check's ID. Hosts that render MCP Apps show it inline; the rest get a URL to give the user as a browser view. If open_app is not among your tools, the MCP server is not registered or the host has not reloaded it: reality-check mcp --help says how to register it. Without it, start reality-check app in the background, since it serves until stopped, and give them the URL it prints or open it automatically if your host supports it.

Keep it current

Keep the tool up to date for the user by default if the permission mode allows it. Each release is a v<version> tag and an install stays on its tag, so uv tool upgrade does not move it. Find the newest release:

git ls-remote --tags --refs --sort=-v:refname https://github.com/microsoft/amplifier-smart-tool-reality-check "v*" | head -1

If it is newer than reality-check --version, run the install command again with that tag and the extras uv tool list --show-extras reports. For a library, run uv add again with the new tag. Update the skill with npx skills add microsoft/amplifier-smart-tool-reality-check#<tag>, adding -g for global installs. Do not try to get around restrictions or operate against the user's preferences to update tools (or skills).

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.