Jira Ticket Enrichment Skill

This skill is triggered when Sentry creates a Jira ticket. It uses MCP tools to:

•Fetch the full Sentry error details (stack trace, tags, context)
•Discover the relevant code files in the GitHub repository
•Analyze the root cause of the error
•Update the Jira ticket with enriched information
•Create a GitHub Pull Request with a detailed fix plan

Trigger

This skill activates when:

•A Jira webhook fires with issue_created event
•The ticket description contains a Sentry Issue link (e.g., JAVASCRIPT-REACT-1)

Cloud & Headless Support

When running in cloud environments (e.g., within Docker or remote workers) without a browser:

•GitHub MCP: Configured via HTTP with a Bearer token (GITHUB_TOKEN) to bypass OAuth browser redirects and Docker-in-Docker requirements.

MCP Tools Required

1. Sentry MCP (`@sentry/mcp-server`)

Used to fetch complete error details from Sentry.

Tools:

•get_sentry_issue: Get issue details including title, status, and metadata
•get_sentry_event: Get full stack trace and context for a specific event

2. GitHub MCP (`@anthropic-ai/github-mcp`)

Used to access the repository, create branches, and open PRs.

Tools:

•get_file_contents: Read specific files from the repository
•search_code: Search for code patterns across the repo
•create_branch: Create a new branch for the fix
•create_pull_request: Open a PR with the plan
•create_or_update_file: Add PLAN.md to the PR

3. Jira API (Direct REST API)

Used to update the Jira ticket with enriched information.

Operations:

•get_issue: Fetch full ticket details
•update_issue: Update description with analysis
•add_comment: Add comments with progress updates

Workflow Steps

Step 0: Notify Jira - Starting Work

Add a comment to the Jira ticket to notify that work has started.

code

Call: jira_api.add_comment(
  issue_key="PROJ-123",
  body="🤖 **AI Agent Started**\n\nThe AI Planning Agent has picked up this ticket and is now:\n- Fetching Sentry error details\n- Analyzing the codebase\n- Creating a fix plan\n\n⏳ This typically takes 2-5 minutes."
)

Step 1: Parse Jira Webhook

Extract the Sentry Issue ID from the ticket description.

code

Input: Jira webhook payload
Extract: Sentry Issue ID (e.g., "JAVASCRIPT-REACT-1")
Extract: Repository name from Sentry tags (e.g., "rom-orlovich/manga-creator")

Step 2: Fetch Sentry Details

Use the Sentry MCP to get the complete error information.

code

Call: mcp_sentry.get_sentry_issue(issue_id="JAVASCRIPT-REACT-1")
Call: mcp_sentry.get_sentry_event(issue_id="JAVASCRIPT-REACT-1", event_id="latest")

Output:
- Full stack trace with all frames
- Error message
- Browser/OS context
- User affected
- Tags (including repository)

Step 3: Discover Relevant Files

Use the GitHub MCP to find and read the affected files.

code

Parse stack trace for file paths:
- /components/Header.tsx:24:17
- /utils/auth.ts:45:3

Call: mcp_github.get_file_contents(repo="rom-orlovich/manga-creator", path="src/components/Header.tsx")
Call: mcp_github.search_code(repo="rom-orlovich/manga-creator", query="onClick Header")

Output:
- Full source code of affected files
- Related files that import/use the affected code

Step 4: Analyze Root Cause

Use Claude's understanding to determine the root cause.

code

Analyze:
- The exact line causing the error
- Why the error occurs (null reference, type error, etc.)
- The broader context of the affected function//component
- Similar patterns in the codebase that might also be affected

Output:
- Root cause explanation
- Affected code scope
- Risk assessment (low/medium/high)

Step 5: Create Fix Plan

Generate a detailed plan following TDD principles.

code

Create PLAN.md with:
1. Problem Summary
2. Root Cause Analysis
3. Affected Files
4. Fix Strategy
5. Test Plan (what tests to write first)
6. Implementation Steps
7. Validation Criteria
8. Risk Assessment

Step 6: Update Jira Ticket

Enrich the ticket with the discovered information.

code

Call: jira_api.update_issue(
  issue_key="PROJ-123",
  description="{enriched_description_with_analysis}"
)

Call: jira_api.add_comment(
  issue_key="PROJ-123", 
  body="🤖 AI Analysis Complete. View the fix plan: {pr_url}"
)

Step 7: Create GitHub PR

Open a draft PR with the plan for human approval.

code

Call: mcp_github.create_branch(repo="...", branch="fix/sentry-JAVASCRIPT-REACT-1")

Call: mcp_github.create_or_update_file(
  repo="...",
  path="PLAN.md",
  content="{plan_content}",
  branch="fix/sentry-JAVASCRIPT-REACT-1"
)

Call: mcp_github.create_pull_request(
  repo="...",
  title="[AI-FIX] Error: This is your first2 error!",
  body="{pr_description_with_plan}",
  head="fix/sentry-JAVASCRIPT-REACT-1",
  base="main",
  draft=true
)

Output Format

Enriched Jira Description

markdown

## 🔍 AI Analysis

### Error Details
**Sentry Issue:** [JAVASCRIPT-REACT-1](https://sentry.io/...)
**First Seen:** 2026-01-19 14:30:00
**Occurrences:** 15
**Users Affected:** 3

### Stack Trace
```javascript
Error: This is your first2 error!
    at onClick (components/Header.tsx:24:17)
    at HTMLButtonElement.handleClick (react-dom.js:123)
    at dispatchEvent (events.js:45)

Root Cause

The error occurs because updateFrom is called on an object that is undefined. This happens when the user clicks the header button before the auth state is initialized.

Affected Files

•src/components/Header.tsx (line 24) - Direct cause
•src/hooks/useAuth.ts - Related auth state management
•src/context/AuthContext.tsx - Provides the auth state

Fix Plan

📄 PR Draft: View Plan

Status: Pending Approval

code


### GitHub PR Description
```markdown
# 🤖 AI-Generated Fix Plan

## Problem
Error `This is your first2 error!` occurring in Header.tsx

## Sentry Issue
[JAVASCRIPT-REACT-1](https://sentry.io/...)

## Root Cause Analysis
The `updateFrom` method is being called on an undefined object because the
auth state has not finished initializing when the user clicks the button.

## Plan

### 1. Write Tests First
- [ ] Add test for button click when auth is loading
- [ ] Add test for button click when auth is ready

### 2. Implement Fix
- [ ] Add null check before calling `updateFrom`
- [ ] Add loading state to prevent premature clicks
- [ ] Update button to be disabled during loading

### 3. Validation
- [ ] All existing tests pass
- [ ] New tests pass
- [ ] Manual test in browser confirms fix

---
⚠️ **This is a Draft PR.** To approve, comment `@agent approve` or use Slack.

Approval Channels

After the skill completes, the task waits for human approval via:

•GitHub: Comment @agent approve on the PR
•Slack: Click the "Approve" button in the notification
•Jira: Transition the ticket to "Approved" status

Configuration

The skill requires these environment variables:

•SENTRY_AUTH_TOKEN: For Sentry MCP authentication
•SENTRY_ORG: Your Sentry organization slug
•GITHUB_TOKEN: For GitHub MCP authentication
•JIRA_URL: Your Atlassian URL
•JIRA_EMAIL: Your Atlassian account email
•JIRA_API_TOKEN: Atlassian API token

Jira Ticket Enrichment Skill

Trigger

Cloud & Headless Support

MCP Tools Required

1. Sentry MCP (@sentry/mcp-server)

2. GitHub MCP (@anthropic-ai/github-mcp)

3. Jira API (Direct REST API)

Workflow Steps

Step 0: Notify Jira - Starting Work

Step 1: Parse Jira Webhook

Step 2: Fetch Sentry Details

Step 3: Discover Relevant Files

Step 4: Analyze Root Cause

Step 5: Create Fix Plan

Step 6: Update Jira Ticket

Step 7: Create GitHub PR

Output Format

Enriched Jira Description

Root Cause

Affected Files

Fix Plan

Approval Channels

Configuration

1. Sentry MCP (`@sentry/mcp-server`)

2. GitHub MCP (`@anthropic-ai/github-mcp`)