← All projects
2026 · Private repository

Jira Ticket Creator

Defects from Excel to Jira incidents, with similar tickets ranked first.

Try it ↓Private repository

What it is

A desktop tool that turns a test-result Excel workbook into Jira incidents. It drives the browser to create each ticket, then links related defects ranked by similarity. It automates the repetitive part of reporting test results.

What it does

  • Restart-safe: the Excel file is the source of truth. Rows that already have an incident number are skipped, so a re-run never creates duplicates.
  • Rows with missing or invalid mandatory fields are skipped and listed in the final summary instead of failing the whole run.
  • A dry-run mode fills the form without clicking Submit.
  • Defect search ranks results with OpenAI Embeddings when a key is configured and falls back to a local model otherwise.
  • Two ways to use it: a desktop chat window that guides the steps, or a classic terminal launcher.
  • If login or one-time-code authentication fails, a debug bundle with URL, visible text and a screenshot is saved automatically.
  • 67 unit tests run without a browser.

How it is built

Python with browser automation for Jira, an Excel reader and writer, an embeddings-based search module and pytest for the test suite. The tool is packaged as patches for distribution to colleagues.

Status

Used by my team. The repository is private because it works against a company Jira instance; the demo below uses invented sample defects.

Stack

  • Python
  • OpenAI Embeddings
  • Playwright
  • pytest

Try it

Runs in your browser on sample data.

This row would be processed and a ticket created.

Similar existing defects

DEF-2101Possible duplicate40%

Navigation route recalculation freezes after leaving a tunnel

DEF-2107Related14%

Climate controls are unresponsive for several seconds after wake from sleep

DEF-2108Related10%

Navigation voice guidance repeats the same instruction twice

Validation follows the rules of the real tool (priority and incident type are mandatory, stability fields are required for stability incidents). Ranking here is TF-IDF cosine similarity on invented sample defects. The real tool uses OpenAI embeddings or a local model on your Jira data.