ContextFlow

Inspiration

Engineering decisions don't live in code alone-they're scattered across Slack conversations, GitHub issues, pull requests, and design discussions. When reviewing an old PR, developers often spend more time searching for context than reviewing code. We built ContextFlow to bridge this gap by making engineering knowledge instantly accessible.

What it does

ContextFlow is an AI-powered Slack application that retrieves the complete context behind any GitHub Pull Request. With a simple /context command, it gathers relevant Slack discussions, GitHub comments, linked issues, summarizes key decisions, and presents slack context to reviewers understand save searching time across several channels.

How we built it

• Slack Bolt (Python) for the Slack application
• GitHub API to retrieve PRs, issues, comments, and timelines
• Google Gemini for reasoning, summarization, and intelligent tool orchestration
• Model Context Protocol (MCP) to expose Slack search as AI tools
• Slack RTS API for context retrieval
• Block Kit

Challenges we ran into

• Working within hackathon time constraints while building an end-to-end production-like workflow.

Accomplishments that we're proud of

• Built a fully functional Slack application that connects Slack discussions with GitHub Pull Requests.
• Implemented AI-powered context retrieval and summarization.
• Delivered an end-to-end working prototype with a polished demo and user experience.

What we learned

This project reinforced that solving engineering problems isn't just about LLMs-it's about retrieval quality, context engineering, tool orchestration, and thoughtful UX. We also gained hands-on experience integrating Slack, GitHub, MCP, and modern AI workflows into a cohesive developer tool.

What's next for ContextFlow

• Support GitHub Issues, commits, and repositories beyond Pull Requests.
• Integrate Notion, Linear, Jira, and Confluence.
• Add semantic search across all engineering knowledge.
• Build organization-wide engineering memory with personalized recommendations and proactive context suggestions.

Built With

• bolt
• githubapi
• google-genai
• mcp
• python
• slack-rts