mobiai-crashlytics

Crashlytics Investigation

Deep investigation of a crash reported through Firebase Crashlytics. Goes beyond a raw stack trace — uses Crashlytics data to understand the full context: user journey, breadcrumbs, affected versions, device info, and event frequency.

When to Use

• User shares a Crashlytics link, crash ID, or issue title
• User asks to investigate a specific Crashlytics issue
• User mentions Firebase crashes or production crash reports from Crashlytics

Prerequisites

This skill requires access to Firebase Crashlytics data. Check what tools are available:

1. Firebase MCP tools — if the Firebase MCP server is configured, use these tools:
   • firebase_update_environment — set the active Firebase project before querying
   • firebase_list_apps — get the app IDs for the project
   • crashlytics_get_issue — fetch issue metadata: event count, user count, error type. Parameters: appId, issueId
   • crashlytics_list_events — fetch crash events with full stack traces. Parameters: appId, issueId, pageSize
2. Firebase CLI — firebase crashlytics:list and related commands
3. User provides the data — if no tools are available, ask the user to paste or export the crash details from the Crashlytics console

If none of these are available, fall back to the mobiai-analyze-crash skill with whatever information the user can provide manually.

Workflow

Step 1: Fetch Crash Details

Get the full crash information from Crashlytics:

• Stack trace — the complete symbolicated stack trace, not just the top frame
• Event count — how many times this crash has occurred
• Affected users — how many unique users have been affected
• Affected versions — which app versions are seeing this crash
• First seen / last seen — when did it start and is it still happening?
• Device breakdown — which devices and OS versions are most affected

Step 2: Analyze the User Journey

Crashlytics provides breadcrumbs — the sequence of events before the crash:

• Screen flow — which screens did the user visit before crashing?
• User actions — taps, navigation events, custom log events
• Network calls — did a network request fail before the crash?
• App state transitions — did the app go to background and come back?
• Custom keys/logs — any custom data the app logs to Crashlytics

This context is what makes Crashlytics investigation more powerful than a raw stack trace. Use it to understand the conditions that trigger the crash.

Step 3: Identify Patterns

Look across multiple crash events for patterns:

• Does it only happen on specific devices or OS versions?
• Does it only happen after a specific user action?
• Did it start with a specific app version? (regression)
• Is it correlated with a specific time of day or network condition?
• Check git log for changes introduced in the first affected version

Step 4: Find the Root Cause in Code

With the full context from Crashlytics, trace the crash in the codebase:

1. Start from the stack trace — find the crashing file and method
2. Use breadcrumbs to understand the state — what screen was the user on? What data was loaded?
3. Check the conditions — what combination of state + action triggers this?
4. Read the code path from the triggering action through to the crash point
5. Check recent changes in the affected files:

   git log --oneline -10 -- <crashing-file>
   

Step 5: Decide — Fix or Report

Based on what you found:

• Clear root cause, simple fix — apply the fix directly. Verify it compiles, run tests.
• Clear root cause, complex fix — explain the root cause, the conditions that trigger it, and propose a fix approach. Include which files need to change and why.
• Unclear root cause — report what you found, what patterns you identified, and what additional information would help narrow it down.

In all cases, include:

• Root cause — what code is wrong and why
• Trigger conditions — what combination of state/actions causes the crash
• Affected scope — which versions, devices, user flows
• Suggested fix — what to change, with risk assessment