Lintbase Mcp
Give your coding agent live Firestore collection and field context so generated queries and types match your real schema.
Overview
Lintbase MCP is a MCP server for the Build phase that injects real-time Firestore schema context into AI coding agents to stop wrong field names in generated code.
What is this MCP server?
- npm package lintbase-mcp (stdio transport) at version 0.1.3
- Pulls real-time Firestore schema context into the agent session
- Targets hallucinated field names and mismatched document shapes in AI-generated code
- Monorepo source at github.com/lintbase/lintbase packages/lintbase-mcp
- stdio MCP fits local Claude Code and Cursor workflows alongside your Firebase project
- npm package lintbase-mcp version 0.1.3
- Transport stdio
- Repository github.com/lintbase/lintbase subfolder packages/lintbase-mcp
Community signal: 1 GitHub stars.
What problem does it solve?
Agents confidently write Firestore code against fields and collections that do not exist in your database.
Who is it for?
Solo builders using Firebase Firestore who want MCP-grounded codegen and refactors without constant manual schema paste-ins.
Skip if: Projects on Postgres-only stacks, teams without Firestore, or builders who only need static OpenAPI stubs with no live database.
What do I get? / Deliverables
After installing lintbase-mcp over stdio, your agent can query live schema context and produce integrations that align with your Firestore documents.
- Agent-accessible live Firestore schema snapshots
- Codegen and queries that reference actual collections and fields
Recommended MCP Servers
Journey fit
Build is where Firestore-backed features are implemented and agents most often invent wrong field names without schema grounding. Integrations is the right shelf because the server bridges your agent to live database structure rather than replacing app UI work.
How it compares
Live Firestore schema MCP bridge, not a generic SQL explorer or a deployment skill.
Common Questions / FAQ
Who is Lintbase MCP for?
Developers using Claude Code, Cursor, or other stdio MCP clients who build on Google Firestore and need agents to respect real collection and field names.
When should I use Lintbase MCP?
Use it during backend and integration work whenever the agent touches Firestore reads, writes, indexes, or TypeScript models derived from your live data model.
How do I add Lintbase MCP to my agent?
Install the npm package lintbase-mcp, configure it as a stdio MCP server in your client, and point it at your Lintbase/Firestore setup following the github.com/lintbase/lintbase packages/lintbase-mcp instructions.