Pipesense

Inspiration

The idea for pipesense was born out of frustration with repetitive manual steps when managing CI/CD pipelines in Bitbucket. As developers, we often waste time switching between the browser, pipeline triggers, and pull requests. I wanted a tool that would not only streamline triggering pipelines but also integrate AI to summarize and review pull requests—helping teams focus on building rather than managing workflows.

What it does

pipesense is a Chrome extension tightly integrated with Bitbucket Pipelines. It allows developers and teams to:

• Trigger pipelines directly from their workflow with minimal friction.
• Automatically review and summarize pull requests using AI, providing quick insights.
• Save time and reduce context switching during CI/CD processes.

How we built it

We built pipesense as a Chrome extension connected to Bitbucket Pipelines APIs. The core pieces include:

• Frontend (TypeScript + React) for the extension UI.
• Bitbucket OAuth integration to authenticate securely.
• AI review engine powered by LLMs to analyze pull requests and summarize changes.
• A lightweight Node.js backend for handling pipeline triggers and AI requests.

We used incremental development—first ensuring reliable pipeline triggers, then layering in AI-powered features.

Challenges we ran into

• Bitbucket API limitations: Handling rate limits and ensuring smooth OAuth flows was tricky.
• Context management for AI: Summarizing pull requests accurately required optimizing prompts and handling large diffs efficiently.
• Chrome Extension quirks: Managing permissions and ensuring a good UX across browsers took careful iteration.
• Performance trade-offs: Balancing AI accuracy with latency and cost was an ongoing challenge.

Accomplishments that we're proud of

• Successfully built a working extension that integrates seamlessly into Bitbucket workflows.
• Created a smooth pipeline trigger flow that saves teams several minutes per task.
• Designed an AI-powered review system that highlights key pull request changes in seconds.
• Established a foundation for future features like customizable rules and advanced analytics.

What we learned

• The importance of prompt engineering and token optimization when working with LLMs.
• How to balance security and usability in OAuth-based extensions.
• That even small workflow improvements in CI/CD can compound into huge productivity gains for teams.
• Building extensions taught us more about browser APIs, background scripts, and content injection than expected.

What's next for pipesense

We plan to expand pipesense by:

• Supporting GitHub Actions and GitLab CI/CD in addition to Bitbucket.
• Adding configurable AI models so teams can choose between cost and accuracy.
• Providing analytics dashboards for pipeline usage and pull request trends.

The journey has just started, but pipesense is already proving to be a valuable assistant for developers and teams aiming to streamline their CI/CD workflows.

Built With

• react
• typescript