GitScribe is a documentation helper for developers who struggle to keep project docs complete and up to date. The core problem is familiar: writing READMEs and change notes often gets postponed, and once the code moves quickly, documentation drifts from what the repository actually does. I built GitScribe to reduce that gap by turning ongoing code changes into documentation updates.
The tool connects to GitHub using an API token, reads repository content and recent commit history, and then generates or refreshes documentation based on what changed. To turn raw code context into readable text, it sends structured prompts to the OpenAI API using an API key. In practice, GitScribe can draft sections like project overviews, setup steps, and “what changed” summaries, and it’s designed to run as an automation so updates can happen regularly across repositories rather than as a one-off manual task.
Building it taught me a lot about end-to-end automation: authenticating to external APIs, handling rate limits and permissions, and shaping prompts so the output stays accurate and consistent. I also learned where the limits are—generated docs still need review, especially when commits are noisy or changes are subtle. Next, I’d like to improve reliability with better diff parsing, clearer source attribution in the output, and optional checks (like linting docs) before committing updates.
