Close the loop from gapto pull request.
The MCP server puts rankings, AI visibility, citations, gaps and briefs inside Claude Code, Codex and OpenCode. Your coding agent reads the gap, generates the brief, drafts the page — and the platform measures the result.
Agents with real context
The MCP tools over streamable HTTP with scoped keys: list projects, query rankings and AI visibility, analyze citations, find content gaps, discover competitors — read tools for context, write tools for briefs and SEO plans.
A reference that can't drift
The tool and route documentation is generated from the running app and gated in CI — regenerate anything else and the build fails. What you read is what ships.
Automations that compose
Pull a growth report into your weekly digest, generate a brief from CI when a new gap appears, sweep audit issues into your issue tracker. Keys are org-scoped; org management stays session-only by design.
Point your agent at the data
One scoped key; configs for Claude Code, Codex and OpenCode in the docs.