🐛 ticket-solver

Jira bug ticket comes in → Claude reads your codebase → solution lands in Slack.

An autonomous agent that listens for Jira webhooks, explores your GitHub repository with AI-powered tool use, and posts a root cause analysis + proposed fix directly to a Slack channel — no human triage needed.

---

How it works

Jira (webhook)
    │
    ▼
ticket-solver (Fastify)
    │  validates label trigger
    ▼
Claude Opus 4.6
    │  iteratively calls tools:
    │  ├── get_file_tree   → explore repo structure
    │  ├── read_file       → inspect relevant source files
    │  └── search_code     → find usages across the codebase
    ▼
Slack
    └── root cause · affected files · proposed fix

The webhook is acknowledged immediately; the analysis runs in the background so Jira never times out.

---

Stack

Layer	Tech
Server	Fastify
AI	Claude Opus 4.6 with adaptive thinking + tool use
GitHub	@octokit/rest
Slack	@slack/web-api
Language	TypeScript (ESM)

---

Getting started

1. Clone & install

git clone https://github.com/joaov56/ticket-solver
cd ticket-solver
npm install

2. Configure environment

cp .env.example .env

Edit .env:

# Jira
TRIGGER_LABEL=ai-solver          # label that activates the agent
JIRA_WEBHOOK_SECRET=             # optional HMAC secret for signature verification
JIRA_BASE_URL=https://your-org.atlassian.net

# Anthropic
ANTHROPIC_API_KEY=sk-ant-...

# GitHub — the repo the agent will read
GITHUB_TOKEN=ghp_...
GITHUB_REPO=owner/repo-name
GITHUB_BRANCH=main

# Slack
SLACK_BOT_TOKEN=xoxb-...
SLACK_CHANNEL=#bugs

3. Run

# development (hot reload)
npm run dev

# production
npm run build && npm start

The server starts on port 3000 (configurable via PORT).

---

Jira setup

1. Go to Jira Settings → System → WebHooks → Create WebHook
2. Set the URL to https://your-server.com/webhook/jira
3. Trigger on: Issue Created and Issue Updated
4. Add the label you configured in TRIGGER_LABEL to any bug ticket

From that point on, every ticket tagged with that label will be automatically analyzed.

---

Slack output

The agent posts a structured message to your channel:

🤖 Bug Analysis: PROJ-123
──────────────────────────────
Ticket:   PROJ-123          Priority: 🔴 High
Reporter: Jane Doe          Labels:   `ai-solver` `backend`

Summary: Payment service throws 500 on retry after timeout

🔍 Root Cause:
The retry logic in PaymentService re-uses the same expired HTTP client
instance instead of creating a new one after a timeout.

📁 Affected Files:
`src/services/payment.ts`, `src/utils/http-client.ts`

💡 Proposed Solution:
Reset the HTTP client instance on timeout in `handleRetry()`:

  if (error instanceof TimeoutError) {
    this.httpClient = new HttpClient(this.config);
  }

Long analyses are posted as thread replies to keep the channel clean.

---

Project structure

src/
├── index.ts              # Fastify server entry point
├── config.ts             # Environment variable validation
├── types/
│   └── index.ts          # Jira webhook & BugTicket types
├── routes/
│   └── webhook.ts        # POST /webhook/jira handler
└── services/
    ├── github.ts         # Repo exploration (list, read, search)
    ├── claude.ts         # Agentic loop with tool use
    └── slack.ts          # Slack message formatting & posting

---

License

MIT