Skip to main content
MCP server CX Codex

pituitary — Claude Code MCP server: drift checks

pituitary checks specs, docs, and code for inconsistencies. Use it when decisions change but related files in the repository may not have caught up.

Open on GitHub

What pituitary does

The server helps identify overlapping decisions in specs, outdated documentation, and code that conflicts with requirements. It can also find old terminology that remains after a team changes its wording. Checks use a shared local index of the corpus, rather than relying only on files in the agent’s context.

The MCP wrapper is an optional way to connect pituitary’s main CLI. The project also provides commands for checking docs and comparing code changes with specs. The core uses a governance SQLite file and a local Stroma snapshot; it does not require Docker or API keys. The MCP wrapper is optional, and the available setup details here do not include a client-specific installation command, so check the selected client’s setup instructions before connecting it.

What it helps with

  • find outdated docs across a repository
  • check a diff against specs before merging
  • spot overlapping decisions in specs
  • find old terminology in documentation

Example prompts for the agent

  • Check the project docs and identify files that have gone stale.
  • Compare this diff with the specs and report conflicts.
  • Find specs that cover overlapping decisions.

How to install pituitary

Install

The MCP registry has no launch command for it: see the repository README

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about pituitary

How do I install pituitary in Claude Code?

The MCP server is an optional wrapper around the CLI. To initialize the CLI on a repository, run pituitary init --path ..

What does pituitary do?

It detects drift between specs, docs, and code, along with overlapping decisions and outdated terminology.

Does pituitary need an API key?

No. The local core uses a governance SQLite file and a local Stroma snapshot, and does not require Docker.

Similar MCP servers

All MCP servers for Writing and docs →

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor