ticket2pr 0.4.2

Automate Jira ticket to GitHub PR workflow

Ticket2PR

Ticket2PR is an AI-powered automation tool designed to streamline the process of converting development tickets into ready-to-merge pull requests.

Workflow

The tool runs a single pipeline from Jira ticket to GitHub PR. Only the steps that truly need AI use it—everything else is done with APIs and deterministic logic. That keeps token usage and run time low.

Full pipeline:

1. Workspace — If no workspace path is set (CLI or config), clone the repo to a temp folder; otherwise use the given path. The temp folder is removed when the run finishes.
2. Fetch Jira issue — Get ticket details via Jira API.
3. Create branch — Generate branch name from issue key/summary, create it on GitHub, and link the branch to the Jira ticket.
4. Checkout branch — Git fetch and checkout.
5. Solve ticket — 2 AI agents in sequence: a Planner explores the codebase and writes a PLAN.md (no other files); an Executor implements that plan (code edits, then stages the changed files).
6. Fix tests (optional, --fix-tests) — 2 AI agents in sequence: a Test planner analyzes staged changes, finds existing related tests, and writes a TESTS_PLAN.md; a Test fixer runs those tests, fixes failures, and stages only the fix-related changes.
7. Pre-commit (if installed and config present) — Run pre-commit; only if it fails, an AI agent tries to fix lint/format issues (with retries). This step is skipped if pre-commit isn’t installed or if .pre-commit-config.yaml is missing or not a file.
8. Commit message & PR body — AI agent: generates a conventional commit message and PR description from the changes and ticket context.
9. Commit and push — Git commit and push.
10. PR title — Build title from issue key and summary, e.g. [PROJ-123] Summary.
11. Create PR — Open the pull request via GitHub API.

Agents

Ticket2PR uses several specialized agents, each with a single responsibility:

Agent	Purpose
Planner	Reads the Jira ticket, explores the codebase, and writes a single PLAN.md with implementation steps and file list. Does not modify any other files.
Executor	Implements the plan: makes the code changes and stages only the files listed in the plan.
Test planner	Analyzes staged changes, finds existing related tests, and writes a TESTS_PLAN.md with test paths and run commands (used only with --fix-tests).
Test fixer	Runs tests from the plan, diagnoses failures, and applies fixes until all tests pass (used only with --fix-tests).
Pre-commit fixer	When pre-commit fails (lint/format, etc.), suggests and applies fixes so hooks pass.
Commit & PR generator	Writes a conventional commit message and the PR body from the diff and ticket context.

They run in sequence where needed; the workflow invokes only the agents required for the current run (e.g. no test fixer without --fix-tests, no pre-commit fixer if hooks pass or pre-commit isn’t available).

Getting Started

Follow these steps to set up and start using Ticket2PR:

Installation

You can install ticket2pr directly from PyPI:

pip install ticket2pr

Usage

Run the Ticket2PR CLI to create a pull request from a Jira ticket:

ticket2pr run <JIRA_ISSUE_KEY>

Replace <JIRA_ISSUE_KEY> with the actual ID of your Jira ticket (e.g., PROJ-123).

Use --fix-tests (or -t) to run an extra step after solving the ticket: the tool will plan and run tests from the staged changes, fix any failures with an AI agent, and stage only the fix-related changes before committing.

Configuration

Ticket2PR automatically guides you through the initial configuration process the first time you run the CLI.

To re-initialize the interactive configuration session, run:

ticket2pr init

Alternatively, you can manually configure settings by editing the ~/.ticket2pr/config.toml file or by setting environment variables.

Prerequisites

• Jira Account and API Token: An account with access to a Jira instance and a valid API token with necessary permissions to view tickets.
• GitHub Account and Personal Access Token: A GitHub account with permissions to create branches and pull requests in your target repository, and a Personal Access Token (PAT) with repo scope for GitHub API access.
• Claude Code: Be logged in locally to Claude Code (used for the AI-powered steps).

License

This project is open-sourced under the terms of the LICENSE.