Jira AI Agent
Automated bug-fix pipeline for VS Code — reads your Jira tickets, writes the fix, tests it, and routes everything through a human Project Manager before anything ships.
How It Works
Jira Ticket → Developer Agent → Tester Agent → PM Review Panel → Git PR
↑ retry loop ↗
- Developer Agent reads the Jira bug ticket + your codebase, identifies the root cause, and writes a targeted fix.
- Tester Agent runs your test suite, evaluates the fix, and writes a regression test. If it fails, it sends structured feedback back to the Developer Agent (up to N retry loops).
- Security Agent (optional) scans the proposed changes for common vulnerabilities.
- PM Review Panel opens with the full diff, test results, AI confidence score, cost estimate, and Approve / Decline / Request Changes actions.
No code ships without your explicit approval.
Features
- 🤖 Three-agent AI pipeline — Developer, Tester, Security Reviewer
- 🔄 Auto-retry loop — configurable max retries before escalating to PM
- 🧠 Feedback memory — learns from past PM decline reasons to improve future fixes
- 💬 Full Jira integration — reads tickets, posts comments, transitions status
- 🌿 Git-aware — creates fix branches, commits changes, builds PR links (GitHub / GitLab / Bitbucket)
- 💰 Cost tracking — estimates token usage and USD cost per pipeline run
- 🔒 Secure credential storage — API keys stored in VS Code SecretStorage, never in settings files
Setup
1. Install the Extension
Install from the VS Code Marketplace or via .vsix file.
2. Open the Setup Dashboard
Press Ctrl+Shift+P → Jira AI Agent: Open Setup Dashboard
Fill in:
| Field | Value |
|---|---|
| Jira URL | https://yourcompany.atlassian.net |
| Jira Email | Your Atlassian account email |
| Jira API Token | Generate at: Atlassian → Account Settings → Security → API tokens |
| AI API Key | Your Google AI (Gemini) or OpenAI API key |
| AI Provider | Gemini (recommended) or OpenAI |
3. Run Your First Pipeline
- Open the Jira AI Agent sidebar panel (robot icon in Activity Bar)
- Your Jira bug tickets will appear in the list
- Click ▶ on any ticket to start the pipeline
Configuration
All settings are available in VS Code Settings (Ctrl+, → search "Jira AI Agent"):
| Setting |
Default |
Description |
jiraAiAgent.jiraUrl |
— |
Jira base URL |
jiraAiAgent.jiraEmail |
— |
Atlassian account email |
jiraAiAgent.jiraProjectKeys |
[] |
Project keys to show (empty = auto-discover) |
jiraAiAgent.aiProvider |
gemini |
gemini or openai |
jiraAiAgent.geminiModel |
gemini-2.5-flash |
Gemini model name |
jiraAiAgent.openAiModel |
gpt-4o |
OpenAI model name |
jiraAiAgent.maxRetries |
3 |
Max Developer↔Tester retry loops |
jiraAiAgent.testCommand |
npm test |
Test suite command |
jiraAiAgent.enableSecurityReview |
true |
Run security scan before PM review |
jiraAiAgent.confidenceThreshold |
0.5 |
Flag low-confidence fixes in PM panel |
jiraAiAgent.protectedBranches |
main, master, develop, release |
Branches the agent cannot modify |
Supported AI Models
Gemini (Google)
gemini-2.5-flash (recommended — fast, cost-effective)gemini-1.5-progemini-1.5-flash
OpenAI
gpt-4o, gpt-4o-mini, gpt-4-turbo, gpt-3.5-turbo
The extension automatically discovers which models are available for your API key and falls back gracefully if a model is deprecated.
Commands
| Command |
Description |
Jira AI Agent: Open Setup Dashboard |
Configure credentials and settings |
Jira AI Agent: Start AI Pipeline |
Run the pipeline on a ticket |
Jira AI Agent: Create Test Bug Ticket |
Create a sample bug ticket in Jira |
Jira AI Agent: Refresh Tickets |
Reload the ticket list from Jira |
Jira AI Agent: View Run History |
Show pipeline run statistics |
Jira AI Agent: Clear Feedback Memory |
Reset learned PM feedback patterns |
Requirements
- VS Code
1.85.0 or later
- A Jira Cloud account with API access
- A Google AI (Gemini) or OpenAI API key
- Git installed and configured in your repository
License
MIT — see LICENSE
