AI Tool Labs

Builds about MCP

MCP servers give an agent tools and memory, but only help if the agent calls them and they return the right thing. These posts measure that with Claude Code: whether a memory server recalls the right note, whether Claude calls a small server at all, which one it uses when a Skill does the same job, and how to load a tool upfront without paying for every tool.

Reading order

  1. How well does an MCP memory server remember? I measured it: how often an MCP memory server returns the right note.
  2. A 97-line MCP server for Claude Code, and when Claude uses it: a 97-line server, and the names and instructions that decide whether Claude calls it.
  3. Claude Code Skill vs MCP server: which one does Claude use?: the same search as a Skill and as an MCP server, and which one Claude picks.
  4. Disable Claude Code tool search? Try MCP alwaysLoad instead: loading one tool upfront with alwaysLoad instead of turning tool search off.

  1. Disable Claude Code tool search? Try MCP alwaysLoad instead

    336 Claude Code runs: one MCP tool marked alwaysLoad was called 96 of 96 times, like tool search off, for 238 tokens of context instead of 12,000.

    Shipped

  2. Claude Code Skill vs MCP server: which one does Claude use?

    I gave Claude Code the same search as a Skill and as an MCP server, 288 runs. It split 25 to 21, and one setting on the server decided it.

    Shipped

  3. A 97-line MCP server for Claude Code, and when Claude uses it

    I built a small MCP server and asked Claude Code 192 questions. A clear tool name got it used in 43 of 48 runs, a vague one in 20. Instructions fixed it.

    Shipped

  4. How well does an MCP memory server remember? I measured it

    24 questions against ai-memory, the MCP memory server I use between sessions: how often the right note comes back, in two languages, and one misfiled page.

    Shipped