AutoTask Engine DTP

AutoTask Engine DTP interface running inside Google Cloud Agent Platform.
Live multi-phase execution showing automated GitHub and Jira telemetry handshakes.
Structured executive markdown report and sprint status summary generation.
Structured executive markdown report and sprint status summary generation.
Architetual Blueprint Diagram

Inspiration

Managing engineering sprints, tracking pull requests, correlating Jira tickets, and dispatching status updates to management typically consumes hours of manual, repetitive operational overhead. For The Taskmaster track, the goal was to eliminate this administrative drag entirely by building an autonomous, intelligent multi-step workflow agent that handles routine DevOps coordination from end to end.

How We Built It

AutoTask Engine DTP was architected and configured natively on Google Cloud's Agent Platform using Gemini 3.5 Flash for advanced reasoning and multi-step execution logic.

Core Agent Platform: Leveraged Agent Studio to build custom system instructions and multi-phase execution loops.
API Handshakes: Simulated secure OAuth/PAT integrations with GitHub and Jira to ingest real-time code telemetry and ticket statuses.
Synthesized Reporting: Processed raw metrics through structured markdown templates to output clean, executive-ready sprint reports and Slack alerts.
Deployment & Version Control: Exported core Python code using the Google GenAI SDK and initialized public/private repositories on GitHub.

Challenges Faced

Designing a reliable autonomous workflow required managing multi-step token limits and orchestrating deterministic tool handshakes (GitHub/Jira to Slack) without human intervention. This was successfully resolved by structuring explicit execution phases and leveraging Gemini 3.5 Flash's robust instruction-following capabilities.

What We Learned

Advanced multi-phase prompt engineering can successfully automate complex, multi-system enterprise workflows.
Combining Google Cloud Agent Platform with the Google GenAI SDK dramatically accelerates the deployment of production-ready autonomous agents.

Future Enhancements

Integrating live webhooks for real-time GitHub PR events.
Deploying the agent pipeline directly onto Google Cloud Run for permanent, zero-touch serverless execution.