Anthropic skill for building MCP servers in Python (FastMCP) or TypeScript, from API research to tests and an eval set.

How it works

Foe: Mirage Djinn, an AI answer that sounds right but isn't.

Foes

  • Mirage Djinn

Party · Context

  • SummonerYour agent100/100
  1. Mirage Djinn

    The Mirage Djinn blocks the way.

    What really happensIt stands for an AI answer that sounds right but isn't.

    Summoner: Context 100/100.

  2. Mirage

    The djinn conjures a mirage, and the hero swings at nothing.

    What really happensA model can sound sure and still be wrong, so AI features need evals and checks.

    Summoner takes 8. Summoner: Context 92/100.

  3. Anthropic MCP Builder

    The builder studies the realm before laying stone.

    What really happensPhase 1 reads the MCP spec, the SDK docs and the target API, then plans which tools to expose and how to name them.

    Mirage Djinn takes 18.

  4. Anthropic MCP Builder

    Tools are carved one by one in Python or TypeScript.

    What really happensPhase 2 sets up the project and implements the tools with FastMCP or the TypeScript MCP SDK.

    Mirage Djinn takes 27.

  5. Mirage

    The Mirage Djinn uses Mirage.

    Summoner takes 8. Summoner: Context 84/100.

  6. Anthropic MCP Builder

    Each stone is tested before the next goes on.

    What really happensPhase 3 reviews code quality, then builds the server and tests it with MCP Inspector.

    Mirage Djinn takes 11. Summoner: Ward.

  7. Anthropic MCP Builder

    Ten hard riddles test whether the tools tell the truth.

    What really happensPhase 4 writes 10 evaluation questions with checkable answers; scripts/evaluation.py runs them against the server.

    Mirage Djinn takes 44. Mirage Djinn falls.

  8. Victory

    The Mirage Djinn falls.

Learn this spell

Claude Code

/plugin marketplace add anthropics/skills
/plugin install example-skills@anthropic-agent-skills

Notes

Four phases: research and planning, implementation, review and test, then writing 10 evaluation questions. Reference files cover MCP best practices and Python and Node guides; scripts/evaluation.py runs the eval set. Claude Code install is the example-skills plugin, which also installs the other example skills. The skill folder carries its own Apache-2.0 LICENSE.txt.

This spell has no recorded sessions in the curator's logs. Tried it? Send a spell report.