Problem solved

Development teams often spend significant time manually picking up issues from trackers, understanding the codebase, and implementing fixes or features. This repetitive context-switching reduces productivity. sortie solves this by automatically converting issue tracker tickets into autonomous coding agent sessions, allowing AI to handle the grunt work of code changes while humans review and approve. It eliminates the manual toil of issue-to-code translation, accelerating development cycles.

Use cases

• Automatically fixing bugs reported in GitHub Issues by having an AI agent analyze, reproduce, and patch the issue.
• Implementing feature requests from Jira by generating code changes and pull requests without human intervention.
• Parallel processing of multiple tickets by running separate agent sessions for each issue simultaneously.
• Using different AI models for different ticket types based on complexity or domain.
• Integrating into CI/CD pipelines to automatically resolve simple tickets and escalate only complex ones.

Audience

• Developers
• ML Engineers
• Data Scientists

Pros

• Agent-agnostic: works with any coding agent that can be invoked from the command line.
• Tracker-agnostic: supports multiple issue trackers via plugins, reducing vendor lock-in.
• Simple deployment: single Go binary with no external dependencies except SQLite, easy to run anywhere.
• Open-source and free to use, with community-driven development.
• Persistent session management with SQLite allows tracking of agent interactions and outcomes.

Cons

• Requires technical knowledge to configure and manage agent integrations.
• Limited tracker support may not cover all issue trackers initially.
• Autonomous code changes may introduce errors and require careful review.