All 94 skillsWeb and data2 of 17
stryx:mcp-server
When building a server that gives Claude new tools (an MCP server).
/stryx:mcp-serverInside its piece: 13 specks, one for each file on its shelf. Its sixteen group-mates glow on the open book behind it. The light splits toward teal, the band of Web and data.
In numbers
counted 2026-10-12 in the installed set- 10reference files
- 1helper scripts
- 13files in the skill
- 187KB on disk
- 6chapters
- 1agent works by this book
Use it when
in the skill’s own words- when designing
- building
- testing
- debugging
- deploying or upgrading an MCP (Model Context Protocol) server that exposes tools
- resources or prompts to LLMs
- "build an MCP server"
- "expose this API to Claude"
- "make a Claude integration"
- "turn this OpenAPI spec or FastAPI app into MCP tools"
- FastMCP (Python) or the MCP TypeScript SDK (@modelcontextprotocol/server or /sdk, McpServer, registerTool)
- Stdio vs Streamable HTTP
- The MCP Inspector
- Tool design and auth
- Evaluating whether a model can use the tools
- Deploying to Prefect Horizon (FastMCP Cloud), Docker or Cloudflare Workers
- errors like "No server object found"
- stdio broken by print or console.log
- transport mismatch
- resource URI
- lifespan
- middleware
- storage or OAuth problems
- FastMCP 2/3 to 4 or TypeScript SDK v1 to v2 upgrades
Not for
another skill does it- Registering an existing server in Claude Code→ stryx:config
- Wiring a service into an app→ stryx:integrations
- ElevenLabs agent MCP tools→ stryx:voice-agents
- Using Mneme→ stryx:mneme
On its shelf
13 files, 187 KB- references/10
- SKILL.md1
- evals/1
- scripts/1
Inside the book
6 chapters, by title only- Choose the stack
- Choose the transport
- The build procedure
- Minimal Python server
- Core rules
- Common errors (full table: [troubleshooting.md](references/troubleshooting.md))
Its helper
1 agent works by this book- stryx:api-engineerWhen an app must talk to an outside API: a typed client, sign-in, retries.
Works with
the skills it names most- stryx:config→When Claude Code's own setup needs checking or tidying: CLAUDE.md, settings, permissions, memory.
- stryx:integrations→When wiring an outside service into an app: Stripe payments, chat messages, API keys.
- stryx:ai-power→When you are not sure which skill fits, or you say "load aipower": it is the map of all skills.
- stryx:mneme→When the Mneme memory server is connected and you ask where code is or what a change breaks.
- stryx:python→When writing or fixing Python 3.11 or newer: types, async code, packaging, pytest.
- stryx:security→When code touches logins, permissions, secrets or encryption, or you ask "is this secure".
Claude loads it by itself when the moment comes; by hand, type the line at the top. It comes with the whole set: Installs all 94 skills, 32 agents and the brain program as one set. The skill’s own text is private source and is not shown here.