Firebase MCP Server
A general-purpose MCP server for Firebase that enables natural language interaction with Firestore, Auth, Storage, and Cloud Functions logs through schema-driven validation and context-efficient tools.
README
Firebase MCP Server
<p> <a href="https://www.npmjs.com/package/@dasasian/firebase-mcp-server"><img alt="npm" src="https://img.shields.io/npm/v/@dasasian/firebase-mcp-server?style=flat-square&color=235a9b"></a> <a href="LICENSE"><img alt="License: MIT" src="https://img.shields.io/badge/license-MIT-235a9b?style=flat-square"></a> <a href="https://modelcontextprotocol.io"><img alt="Model Context Protocol server" src="https://img.shields.io/badge/MCP-server-235a9b?style=flat-square"></a> <img alt="Node 18+" src="https://img.shields.io/badge/node-18%2B-5fa04e?style=flat-square"> </p>
General-purpose Model Context Protocol (MCP) server for Firebase (Firestore, Storage, Auth, Functions Logging) with schema-driven validation and context-efficient tools.
Features
- @ Mention Support - Reference Firestore documents with
@firebase:firestore://users/user-123 - Smart Autocomplete - MRU cache tracks accessed documents for quick re-reference
- Auto-Discovery - Shows both schema-based AND discovered collections
- Path-based schemas - Follows Firebase
firestore.rulesconvention - Schema evolution - Field status metadata (experimental → official → legacy)
- Hot reload - File watching, no restart needed when schemas change
- Flexible validation - Three modes: strict, warn (default), permissive
- Works without schemas - Discovery mode for exploring unknown databases
- Context-efficient tools - 99% token reduction for large datasets
- Index-aware queries - Validates queries against
firestore.indexes.json - Functions logging - SQL-like queries for Cloud Functions logs with aggregations and label filtering
Install
Add it to your MCP client (e.g. Claude Code) — runs via npx, no global install needed:
{
"mcpServers": {
"firebase": {
"command": "npx",
"args": ["-y", "@dasasian/firebase-mcp-server", "start", "./firestore-schemas.json"]
}
}
}
start takes an optional schema config path (default ./firestore-schemas.json) and an optional indexes path (default ./firestore.indexes.json) — see Schema Format. Firebase auth uses Application Default Credentials.
Or install the CLI globally:
npm install -g @dasasian/firebase-mcp-server
firebase-mcp start ./firestore-schemas.json
Development (from source)
npm install
npm run build
npm run cli -- start --config ./examples/basic/firestore-schemas.json
@ Mention Support (Resources)
Reference Firestore documents directly in Claude Code:
What's the email for @firebase:firestore://users/user-123?
Show me all posts: @firebase:firestore://posts/*
How It Works
When you type @firebase in Claude Code:
Shows schema-based collections (with validation):
- 📋 Users (User account documents)
- 📋 Posts (Blog post documents)
- 📋 Comments (Comments on posts)
Plus auto-discovered collections (no schema):
- 🔍 analytics
- 🔍 sessions
- 🔍 audit_logs
Configuration
# Enable/disable auto-discovery (default: true)
export FIRESTORE_AUTO_DISCOVER=true
# Cache duration in seconds (default: 300 = 5 minutes)
export FIRESTORE_DISCOVERY_CACHE_TTL=300
Auto-discovery cost: ~$0.0003/day (negligible)
Tools
The server ships 33 tools in four groups. Every tool declares MCP annotations
(readOnlyHint, destructiveHint, idempotentHint, openWorldHint) so a client
can auto-approve reads and ask before writes.
Choosing which tools to load
All 33 tool definitions cost roughly 12k tokens of context on every session. If you only need part of the surface, narrow it:
# Only Firestore (12 tools, ~4.3k tokens)
firebase-mcp start --tools firestore
# Firestore plus Auth
firebase-mcp start --tools firestore,auth
Or set it in your MCP client config:
export FIREBASE_MCP_TOOLS=firestore,storage
The --tools flag wins over the environment variable. Leaving both unset loads
everything, so upgrading never hides a tool you were already using. Calling a
tool from a group you switched off returns an error naming the group to add.
| Group | Tools | ~Tokens |
|---|---|---|
firestore |
12 | 4,341 |
storage |
14 | 4,152 |
auth |
6 | 2,075 |
logs |
1 | 1,817 |
| all (default) | 33 | 12,387 |
Firestore (12) — --tools firestore
firestore_show_collections— Show collections (read)firestore_read— Read document (read)firestore_export— Export collection (read)firestore_validate— Validate against schema (read)firestore_query_select— Query documents (read)firestore_query_collection_group— Query collection group (read)firestore_count— Count documents (read)firestore_sum— Sum a field (read)firestore_stats— Collection statistics (read)firestore_import— Import document (write, destructive)firestore_update— Update documents (write, destructive)firestore_delete— Delete documents (write, destructive)
Firebase Auth (6) — --tools auth
firebase_auth_list_users— List users (read)firebase_auth_get_user— Get user (read)firebase_auth_create_user— Create user (write)firebase_auth_update_user— Update user (write, destructive)firebase_auth_delete_user— Delete user (write, destructive)firebase_auth_revoke_sessions— Revoke sessions (write, destructive)
Firebase Storage (14) — --tools storage
firebase_storage_list_buckets— List buckets (read)firebase_storage_ls— List files (read)firebase_storage_stat— File metadata (read)firebase_storage_find— Find files (read)firebase_storage_get_url— Get file URL (read)firebase_storage_get_access— Get file access (read)firebase_storage_read— Download file to a local temp path (write)firebase_storage_upload— Upload file (write, destructive)firebase_storage_rm— Delete file (write, destructive)firebase_storage_cp— Copy file (write, destructive)firebase_storage_mv— Move file (write, destructive)firebase_storage_sync— Sync bucket to local (write, destructive)firebase_storage_push— Push local to bucket (write, destructive)firebase_storage_set_access— Set file access (write, destructive)
Cloud Logging (1) — --tools logs
firebase_functions_logs— Query Cloud Functions logs with SQL-like syntax (write)
firebase_storage_read and firebase_functions_logs are not marked read-only
because they write: the first downloads to a local temp file, the second updates
its auto-discovered logging schema on disk.
Reading function logs needs an extra IAM role. The Firebase Admin SDK service account has no Cloud Logging access by default, so
firebase_functions_logsfails withPERMISSION_DENIED: Permission denied for all log viewseven when every other tool works. Grant the role and allow a few minutes for it to take effect:gcloud projects add-iam-policy-binding PROJECT_ID \ --member="serviceAccount:firebase-adminsdk-xxxxx@PROJECT_ID.iam.gserviceaccount.com" \ --role="roles/logging.viewAccessor"
roles/logging.vieweralone may not be enough — the error names log views, and thelogging.views.accesspermission is inviewAccessor.
Example queries:
// Discover what functions exist
{"distinct": "functionName"}
// Show recent errors
{"where": [{"field": "severity", "operator": "==", "value": "ERROR"}], "limit": 20}
// Top error patterns with counts (use "message" — structured logs have no textPayload)
{"groupBy": ["message"], "aggregates": [{"field": "*", "operation": "count", "alias": "count"}], "where": [{"field": "severity", "operator": "==", "value": "ERROR"}], "orderBy": [{"field": "count", "direction": "desc"}], "limit": 10}
// Filter by custom labels (e.g., user, environment)
{"where": [{"field": "labels.user_id", "operator": "==", "value": "123"}]}
Schema Format
Schemas follow Firebase's path-based convention:
{
"schemas": {
"/organizations/{organizationId}": {
"description": "Organization documents",
"schema": {
"type": "object",
"required": ["id", "name"],
"properties": {
"id": { "type": "string" },
"name": { "type": "string" }
}
}
},
"/organizations/{organizationId}/products/{productId}": {
"description": "Product catalog",
"schema": {
"type": "object",
"properties": {
"name": { "type": "string" },
"category": {
"type": "string",
"x-status": "legacy",
"x-replacedBy": "productType"
},
"productType": {
"type": "string",
"x-status": "experimental"
}
}
},
"timestampFields": ["createdAt", "updatedAt"]
}
}
}
Documentation
- Installation Guide
- @ Mention Support Guide - How to use
@firebase:firestore://references - Configuration Reference
- Schema Creation Guide
- Functions Logging Guide - Query Cloud Functions logs with SQL-like syntax, aggregations, and custom labels
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.