RetentaRetenta

Evidence record · captured 2026-10-11

MCP Server Builder

A four-phase process for building MCP servers in Python (FastMCP) or TypeScript (MCP SDK): research the target API and MCP best practices, implement tools with good naming, schemas and error messages, review and test, then write an evaluation of realistic questions. Reference files cover MCP best practices, per-language server guides and how to build evaluations.

Official publisherUnverifiedSource reviewedBehavior not tested
Source review

Editorial assessment

This is a source-only assessment. It does not claim runtime or security testing.

What it does

Turns MCP server work into a four-phase process that ends with ten read-only, string-checkable eval questions, so a server is judged by whether a model can use its tools, not by whether it compiles.

Best for

  • Wrapping an existing REST API as an MCP server in TypeScript or Python
  • Reviewing tool names, input schemas and annotations on a server before release
  • Writing a small QA set to check that a model can actually use a server's tools

Strengths in the source

  • Gives concrete naming rules, such as service-prefixed snake_case tools like github_create_issue, so tools stay distinguishable when several servers are loaded at once.
  • Has the agent set readOnlyHint, destructiveHint, idempotentHint and openWorldHint on each tool, and define outputSchema with structuredContent where the SDK supports it.
  • Requires each of the ten eval questions to be independent, read-only, stable over time and checkable by string comparison, with the agent solving each one itself first.
  • Its best-practices reference covers local streamable HTTP hardening: validate the Origin header and bind to 127.0.0.1 rather than 0.0.0.0.

Limitations

  • Running the bundled evaluation.py harness needs an ANTHROPIC_API_KEY and calls the Claude API, so every eval run is billed to that key.
  • Fetches the MCP TypeScript and Python SDK READMEs from their main branches at run time, so part of the guidance is whatever those files say that day, not the pinned copy.
  • The evaluation guide at this commit names claude-3-7-sonnet-20250219 as the harness's default model, a dated ID worth overriding with -m before a run.

When not to use it

  • You only need to connect to or call an existing MCP server, not build one.
  • A throwaway personal server where a ten-question evaluation set is more ceremony than the job needs.

Reviewed by Retenta editorial on 2026-10-11; scope: source only.

Files included in this review
  • skills/mcp-builder/SKILL.md

Evidence and boundaries

Execution · not tested

Retenta has not executed this skill.

Security · not tested

Bundles 2 scripts (scripts/evaluation.py, scripts/connections.py) that the agent may run. Retenta has not run them or scanned them.

Pinned publisher source
skills/mcp-builder/SKILL.md
Revision
dbd4588f9e1033efb41dad4bef2f7947c8993d44
License evidence
Apache-2.0Retenta preserves the license text captured with this immutable upstream snapshot; no independent legal or security review is claimed.
Last source change
2025-12-01
Independent verification
Published by Anthropic, but not independently verified by a third-party reviewer.
Upstream repository
Open publisher repository

Compatibility

Claude Code (plugin marketplace)

documented

Named in the upstream skill or repository documentation; not tested by Retenta.

Claude.ai

documented

Named in the upstream skill or repository documentation; not tested by Retenta.

Claude API (Skills API)

documented

Named in the upstream skill or repository documentation; not tested by Retenta.

Requirements and access

  • Python (mcp, anthropic packages) or Node.js MCP SDK
  • ANTHROPIC_API_KEY to run the bundled evaluation harness
  • Network access to the target API and MCP docs

Referenced tools

No named tool is required by the reviewed SKILL.md.

Install from the source

Clones the current default branch, which may be newer than the reviewed revision dbd4588. The destination shown is Claude Code's personal skills folder; other agents read the same folder from their own skills directory.

View the SKILL.md at its source Verify the publisher's install documentation

More AI development skills →

Install the full skill folder

git clone --depth 1 https://github.com/anthropics/skills.git
cp -R skills/skills/mcp-builder ~/.claude/skills/mcp-builder

Thyme RAG Capsules

Make this skill’s state durable, governed, and portable.

A Retenta capsule lets agents recall prior work, append provenance-stamped progress, and carry the memory to the next run.

Create durable agent memoryUnderstand the evidence labels