Pituitary
Detect when specs, docs, and code diverge so your agent-assisted project does not ship outdated contracts or duplicate modules.
Overview
Pituitary is a MCP server for the Ship phase that detects spec drift, stale docs, overlapping code, and documentation–implementation mismatch.
What is this MCP server?
- Flags spec drift, overlapping modules, stale documentation, and implementation drift before they compound.
- Ships as mcpb binaries for darwin-arm64 and linux-amd64 (v1.0.0-beta.8) with stdio transport.
- From Dusk Network—aimed at teams where docs and specs must stay aligned with code under fast iteration.
- Agent-callable drift detection for solo builders juggling SKILL.md, OpenAPI, and README truth.
- Beta release channel (1.0.0-beta.8) with pinned file SHA256 on registry packages.
- Version 1.0.0-beta.8 with two mcpb packages (darwin-arm64, linux-amd64).
- Transport: stdio via registryType mcpb.
- Publisher focus: overlap, stale docs, and code drift detection.
Community signal: 21 GitHub stars.
What problem does it solve?
Fast agent edits leave specs and docs lying about what the code actually does, and you only discover the gap when integrations break.
Who is it for?
Builders maintaining specs, OpenAPI, or design docs alongside code who need automated drift radar in the agent loop.
Skip if: Greenfield spikes with no written spec yet, or teams that want runtime error tracking instead of documentation alignment.
What do I get? / Deliverables
After adding Pituitary MCP, your agent can surface drift and overlap early so you reconcile docs and specs before customers rely on them.
- Reports on overlap, stale docs, and spec versus code drift.
- Action list to reconcile documentation before release.
- Ongoing drift signal during operate-phase iteration.
Recommended MCP Servers
Journey fit
Catching spec drift and stale documentation is a review discipline that prevents bad launches and keeps operate iterations honest. Pituitary targets overlap, stale docs, and code–spec drift—classic human and agent review gates, not net-new feature coding.
How it compares
Spec-and-doc drift MCP, not application security scanning or CI deploy orchestration.
Common Questions / FAQ
Who is io.github.dusk-network/pituitary for?
Solo and small teams on spec-driven or doc-heavy projects who use Claude Code or Cursor and need drift detection without a separate doc platform.
When should I use io.github.dusk-network/pituitary?
Use it before release review, after large refactors, and whenever agents rewrite modules without updating specs or README sections.
How do I add io.github.dusk-network/pituitary to my agent?
Download the matching pituitary mcpb release (v1.0.0-beta.8) for darwin-arm64 or linux-amd64 and register it as a stdio MCP server in your host config.