Firebase
Auth, Firestore, Functions.
Verdict
Common use cases
- Query Firestore collections during standups
- Check user authentication status in support threads
- Pull real-time database snapshots for debugging
- Audit security rules before deployment
- Export usage metrics for weekly reports
Integration
- Vendor
- Firebase
- Category
- developer-tools
- Auth
- OAUTH2
- Composio slug
firebase
Tools
Setup
Setup guide
- 11. In Switchy, open your Space settings and click 'Add Integration'. 2. Search for Firebase and select it from the developer tools category. 3. Click 'Connect' to start the OAuth flow in a new browser tab. 4. Sign in with the Google account that owns your Firebase project. 5. Select the Firebase project you want Switchy to access. 6. Review the requested scopes—typically read access to Firestore, Realtime Database, and Authentication—then click 'Allow'. 7. You'll return to Switchy with a green confirmation banner. 8. Inside any Space, type '@Firebase list collections' to verify the connection works. 9. If you see your Firestore collections listed, setup is complete. 10. To grant write access later, revisit Space settings and click 'Manage Permissions' next to Firebase.
What teammates see: by default, memories from Firebase are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
List Active Users
@Firebase show me all users who signed in during the past 24 hours, grouped by authentication providerOpen in a Space →
Query Recent Orders
@Firebase get the 20 most recent documents from the orders collection, sorted by timestamp descendingOpen in a Space →
Check Database Rules
@Firebase show me the current security rules for my Firestore database and highlight any public read accessOpen in a Space →
Monitor Collection Size
@Firebase count the total documents in the users, sessions, and events collections and compare to last weekOpen in a Space →
Inspect User Record
@Firebase find the user with email support@example.com and show their UID, creation date, and custom claimsOpen in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Firebase show me the top 10 most active users in my production Firestore collection over the last 7 days
Based on your Firestore analytics, here are the 10 users with the highest activity counts in the past week: user_8f3a2 (247 events), user_6d91c (189 events), user_2b4e7 (156 events), and seven others ranging from 98 to 134 events. Activity includes document reads, writes, and real-time listener connections. This data reflects your current billing period and may help identify power users or potential optimization targets.
This example assumes the MCP can query Firestore usage metrics or Analytics data. Requires OAuth scopes for both Firestore and Firebase Analytics. Response time depends on collection size and indexing — large datasets may hit query timeouts or require pagination.
@Firebase create a new Cloud Function triggered by writes to the /orders collection that sends a confirmation email via SendGrid
I've drafted a Cloud Function deployment configuration for your orders collection trigger. The function will execute on document creation in /orders, extract the customer email and order ID, and call SendGrid's API with your template. You'll need to deploy this via Firebase CLI with 'firebase deploy --only functions:sendOrderConfirmation' and set your SENDGRID_API_KEY in environment config. Estimated cold-start latency: 800ms; warm invocations under 200ms.
This example shows function scaffolding and deployment guidance, not live deployment — the MCP cannot execute CLI commands or modify your Firebase project directly without explicit write permissions. You must review generated code and deploy manually. Requires OAuth scopes for Cloud Functions management.
@Firebase analyze my Authentication sign-in methods and recommend which providers to prioritize based on last month's conversion rates
Your Firebase Authentication data shows Google Sign-In has an 87% completion rate (4,231 successful logins from 4,862 attempts), while Email/Password sits at 62% (1,890 from 3,048 attempts). Anonymous sign-in converts at 94% but represents only 8% of total volume. Recommendation: prioritize Google Sign-In in your UI flow and consider adding Apple Sign-In for iOS users, which typically sees 80%+ conversion. Email/Password's lower rate suggests friction in your password reset or validation flow.
This synthesis example combines Authentication logs with the AI's pattern recognition to surface actionable insights. Accuracy depends on how your Firebase project tracks sign-in events — incomplete logging or custom auth flows may skew results. Requires OAuth scopes for Authentication and Analytics read access.
Use-case deep-dives
When Firebase MCP helps debug production issues faster
A 3-person mobile team ships iOS and Android apps with 20k active users. Crashes spike after a release, and the on-call engineer needs to pull Crashlytics logs, check recent Remote Config changes, and correlate with Analytics events—all while the product lead asks for an ETA in Slack. The Firebase MCP lets you query crash reports, read config flags, and pull event funnels without leaving the AI workspace. If your team already lives in Firebase Console for these tasks, this MCP saves 4-6 tab switches per incident. The trade-off: OAuth2 setup requires a service account with read scopes across Crashlytics, Analytics, and Remote Config. If you only check Firebase once a week, the auth overhead isn't worth it. But if you triage mobile issues daily, this MCP turns your AI into a Firebase power-user.
Pulling user context from Firestore during support calls
A 5-person SaaS team uses Firestore as their primary database. Support tickets come in via Intercom, and the rep needs to check the user's subscription status, recent activity logs, and feature flags before replying. Without the Firebase MCP, this means opening Firestore Console, navigating collections, and copying JSON back into the ticket. With the MCP, the AI pulls the user document, parses nested fields, and surfaces the answer in one prompt. This works if your Firestore schema is under 50 collections and your support team handles 10+ tickets a day. If your database is heavily normalized or you use a different backend, this MCP won't help. But for Firestore-native teams, it cuts ticket resolution time by 30-40%.
When Firebase MCP speeds up experiment review meetings
A 6-person growth team runs weekly A/B tests via Firebase Remote Config and tracks conversions in Google Analytics. Every Friday, the PM asks: 'Did variant B hit the 5% lift target?' The engineer opens Firebase Console, exports CSV data, and pastes it into a spreadsheet. The Firebase MCP automates this: query the experiment results, calculate confidence intervals, and summarize the outcome in plain English. This saves 15 minutes per experiment if you run 3+ tests a month. The catch: OAuth2 requires read access to both Remote Config and Analytics, and if your experiments live in Optimizely or LaunchDarkly instead, this MCP is irrelevant. For teams all-in on Firebase, it turns experiment review from a chore into a one-prompt task.
Frequently asked
What does the Firebase MCP let me do in Switchy?
It connects your Firebase project so AI agents can read analytics, query Firestore collections, check authentication logs, and inspect Cloud Functions. You're not writing code or clicking through the console — you ask questions in natural language and the MCP fetches the data. Useful for debugging user flows, auditing security rules, or pulling metrics into a shared workspace.
Do I need Firebase admin rights to connect this MCP?
Yes. The OAuth flow requires a Google account with at least Editor or Owner permissions on the Firebase project. Viewer access won't work because the MCP needs to read configuration and query backend services. If you're on a team plan, only one person needs to authenticate — everyone else inherits read access through Switchy's workspace permissions.
Can it deploy Cloud Functions or update Firestore documents?
No. This MCP is read-only by design. It pulls data from Firebase services but won't push changes, deploy code, or modify security rules. If you need to write data, use the Firebase CLI or console directly. The MCP exists to answer questions and surface insights, not to replace your deployment pipeline.
Why use this instead of just opening the Firebase console?
The console makes you click through tabs and remember where things live. The MCP lets you ask "how many users signed up this week" or "which function is throwing errors" and get answers in seconds. It's faster for ad-hoc questions and keeps the context in your Switchy workspace, so the whole team sees the same data without learning Firebase's UI.
Does connecting Firebase count against my Switchy plan limits?
The connection itself is free — it's one MCP slot. Usage counts against your monthly message quota because every query the AI makes to Firebase burns tokens. Heavy analytics pulls or large Firestore scans will use more quota than simple auth checks. If you hit limits, upgrade your plan or disconnect MCPs you're not actively using.