agentleFS
Sign inStart free
Wsh7AshWsh7Ash/skillmcp/skills/server-discovery-and-trust/SKILL.md

server-discovery-and-trust

Skill0 starsChanged 15 days agoMITView on GitHub

Decide which third-party tool servers to load by reading what they do before granting them a call surface. Use when an agent is about to be pointed at a server you did not write

  • Installs packages

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the tool list and every description before enabling the server

    line 21
  2. Distinctive:

    Check what the server can reach: filesystem paths, network hosts, credentials

    line 22
  3. Distinctive:

    Prefer servers that read local files over ones that proxy network access

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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
SKILL.md

Server Discovery and Trust

Decide which third-party tool servers to load by reading what they do before granting them a call surface

Use when: an agent is about to be pointed at a server you did not write

Avoid when: the server is internal, versioned and already reviewed by your own team

Procedure

  1. Identify the publisher and the exact version or commit being installed
  2. Read the tool list and every description before enabling the server
  3. Check what the server can reach: filesystem paths, network hosts, credentials
  4. Prefer servers that read local files over ones that proxy network access
  5. Pin the version so an upstream change cannot alter behaviour silently
  6. Record the review in the project so the next person inherits the decision

Signals

  • the server's tools are read-only or clearly bounded
  • the package is pinned to a commit rather than a floating tag
  • the server requests no credentials it does not strictly need
  • tool descriptions state what is read and what is written

Patterns

1. Pinned install

pip install skillmcp==1.0.0

2. Surface review

list_packs, search_skills, get_skill: read-only

3. Reach check

reads a local packs directory, no network

Pitfalls

  • Loading a server because its description sounds useful without reading its tools
  • Granting filesystem access to a server that only needs one directory
  • Trusting a floating tag that could be repointed at any time

Done when

  • the tool list of every third-party server has been read and recorded
  • each server is pinned to a version or commit

References

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.

Check your setup, free