Sentry Issue Fixer

Automatically fetch the most important open Sentry issue, analyze it, implement a fix, verify with tests, and create a pull request.

Prerequisites

• Sentry MCP server connected (configured in

  ~/.claude/settings.json

  )
• GitHub CLI (

  gh

  ) authenticated
• Docker running (for integration tests)
• Python 3.12 with virtualenv
• Checkout

  main

  branch and pull the latest changes

MCP Server Configuration

The Sentry MCP server should be configured in

~/.claude/settings.json

{
  "mcpServers": {
    "Sentry": {
      "url": "https://mcp.sentry.dev/mcp"
    }
  }
}

On first use, you'll be prompted to authenticate with Sentry via OAuth.

Workflow

Step 1: Fetch Most Important Open Issue

Use Sentry MCP tools to get the highest priority unresolved issue:

1. First, list available organizations:
   mcp__Sentry__list_organizations

2. List projects in the organization:
   mcp__Sentry__list_projects with organization_slug

3. Search for unresolved issues sorted by priority/frequency:
   mcp__Sentry__search_issues with:
   - organization_slug
   - project_slug
   - query: "is:unresolved"
   - sort: "freq" or "priority"

4. Get detailed issue information:
   mcp__Sentry__get_issue with issue_id

Step 2: Analyze the Issue

Use Sentry's analysis tools:

1. Get issue details and stack trace:
   mcp__Sentry__get_issue_details

2. Find errors in specific files:
   mcp__Sentry__find_errors_in_file with filename from stack trace

3. Use Seer AI for root cause analysis (if available):
   mcp__Sentry__invoke_seer_agent for AI-powered fix suggestions

Gather from analysis:

• Exception type and message
• Stack trace (file, function, line number)
• Error frequency and affected users
• Environment context (tags, release, etc.)
• Pattern of occurrences

Step 3: Plan the Fix

Before implementing, create a plan:

1. Root Cause Analysis

   • What is causing the error?
   • Is it a logic error, missing validation, race condition, etc.?

2. Proposed Solution

   • What changes are needed?
   • Which files need modification?
   • Are there related issues to consider?

3. Risk Assessment

   • Could this fix break other functionality?
   • Does it need backward compatibility?

4. Testing Strategy

   • What unit tests should be added/modified?
   • What integration tests are relevant?

Step 4: Implement the Fix

1. Create a Feature Branch

   git checkout main
   git pull origin main
   git checkout -b fix/sentry-<issue-id>-<short-description>
   

2. Make Code Changes

   • Implement the planned fix
   • Add appropriate error handling
   • Add or update tests

3. Commit Changes

   git add <files>
   git commit -m "fix: <description>
   
   Fixes Sentry issue <issue-id>
   
   Co-Authored-By: Claude <noreply@anthropic.com>"
   

Step 5: Run ALL Test Suites

IMPORTANT: You MUST run ALL of the following test suites before creating a PR. Do NOT skip any.

5.1 DB/SQLAlchemy Tests (Primary Backend Tests - Dockerized)

docker compose -f tests/db-sqlalchemy/docker-compose.yml --project-directory . run --build --rm tests

5.2 Backend Unit Tests

.venv/bin/pytest tests/unit/ -v --tb=short --ignore=tests/unit/test_rpc_endpoint_manager.py

5.3 Frontend Unit Tests

cd frontend && npm run test

If any tests fail:

• Analyze the failure
• Fix the issue
• Re-run ALL test suites until they pass

Step 6: Run Integration Tests (Optional - if Docker services are running)

Follow the integration-tests skill:

# Ensure Docker is running with the studio
docker compose ps  # Verify services are up

source .venv/bin/activate
export PYTHONPATH="$(pwd)"

# Run integration tests in parallel (excluding test_validators.py)
gltest --contracts-dir . tests/integration -n 4 --ignore=tests/integration/test_validators.py

# Run validator tests separately
gltest --contracts-dir . tests/integration/test_validators.py

If tests fail:

• Check Docker logs:

  docker compose logs -f

• Analyze test output
• Fix issues and re-run

Step 7: Create Pull Request

Follow the create-pr skill:

# Push branch
git push -u origin $(git branch --show-current)

# Create PR with Sentry context
gh pr create --title "fix: <description from Sentry issue>" --body "$(cat <<'EOF'
Fixes Sentry issue: <link-to-sentry-issue>

# What

- [Describe the error that was occurring]
- [List the changes made to fix it]

# Why

- This error was affecting [X users / occurring Y times]
- Root cause: [explanation]

# Testing done

- [x] Unit tests pass
- [x] Integration tests pass
- [x] Verified fix resolves the Sentry error pattern

# Decisions made

- [Document any non-obvious choices]

# Checks

- [x] I have tested this code
- [x] I have reviewed my own PR
- [x] I have set a descriptive PR title compliant with conventional commits

# Reviewing tips

- Focus on [specific areas]
- The main change is in [file/function]

# User facing release notes

- Fixed: [user-visible description of what was broken]
EOF
)"

Sentry MCP Tools Reference

Based on Sentry MCP documentation (https://docs.sentry.io/product/sentry-mcp/):

Common Tool Patterns

# List organizations you have access to
mcp__Sentry__list_organizations

# List projects in an organization
mcp__Sentry__list_projects(organization_slug)

# Search for issues
mcp__Sentry__search_issues(organization_slug, query="is:unresolved")

# Get issue details
mcp__Sentry__get_issue(issue_id)

# Find errors in a specific file
mcp__Sentry__find_errors_in_file(organization_slug, project_slug, filename)

# Use Seer AI agent for analysis
mcp__Sentry__invoke_seer(issue_id)

Common Issue Patterns

NoneType

KeyError

.get()

Example Session

1. List organizations:
   > mcp__Sentry__list_organizations
   Result: [{"slug": "genlayer", ...}]

2. Search unresolved issues:
   > mcp__Sentry__search_issues(org="genlayer", query="is:unresolved", sort="freq")
   Result: Top issue: "KeyError: 'validator_address' in consensus/worker.py"

3. Get issue details:
   > mcp__Sentry__get_issue(issue_id="12345")
   Result: Stack trace showing error at worker.py:145

4. Analyze:
   - Error in worker.py:145 accessing result['validator_address']
   - Occurs when validator response is incomplete
   - 47 events, affecting 12 users

5. Plan:
   - Add validation before accessing key
   - Log warning for incomplete responses
   - Add unit test for edge case

6. Implement:
   - Edit worker.py to add .get() with default
   - Add test_worker_missing_validator_address test

7. Test:
   - gltest --contracts-dir . tests/unit/test_worker*.py
   - All pass

8. Create PR:
   - gh pr create with Sentry context

Troubleshooting

Sentry MCP Not Connected

• Restart Claude Code to trigger OAuth flow
• Check

  ~/.claude/settings.json

  has the mcpServers configuration
• Verify you have access to the Sentry organization

Cannot Find Issues

• Check organization and project slugs are correct
• Verify you have the right permissions in Sentry
• Try different query filters (is:unresolved, is:unhandled)

Cannot Reproduce Issue Locally

• Check environment differences (env vars, config)
• Use Sentry event data to understand exact conditions
• Add logging to capture more context

Tests Timeout

• Use

  --leader-only

  for faster integration tests
• Run specific test files instead of full suite