Now live
11GitHub & Bitbucket
Connect repos from both providers. Webhooks registered automatically on connect.
AI docs on every push
Every commit triggers documentation generation for changed files only.
Bring your own API key
Groq, Gemini, Anthropic, OpenAI, Mistral. Switch anytime, zero lock-in.
Smart regeneration
Skip pushes that only change whitespace, comments, or imports. No AI cost.
Semantic regeneration
A lightweight model checks whether the diff actually affects the docs before triggering a full update.
Exclusion list
Define glob patterns to skip folders, file types, or generated code.
Pause / resume repos
Stop doc updates during rapid releases, resume when you're ready.
Export as Markdown ZIP
Download all documentation as .md files, ready for Docusaurus, VitePress, or MkDocs.
Edit docs in the browser
Fix or extend any doc directly in the UI. No commits needed.
Job queue with live view
See processing status in real time. Retry, stop, or clear jobs individually.
Pretty formatting
One-click normalization of AI output: headings, bullets, spacing. Without touching the content.
Coming soon
6GitLab
Same integration model as Bitbucket: OAuth token + webhook HMAC.
File chunking (413)
Split files that exceed a model's context window and merge the results.
Multi-provider AI
Configure multiple API keys simultaneously and route by repo or file type.
Provider fallback
Automatically switch to a backup provider when the primary hits a rate limit or daily cap.
Ollama
Fully local, zero-cost, zero-privacy-concern generation. Run any open model on your own hardware.
Public API
Call Readmify programmatically: pass your AI provider, API key, file content or repo reference, and get documentation back. Useful for CI pipelines, custom agents, and third-party integrations.
Changelog
2026-07-12
- —Semantic regeneration mode: cheap model pre-check before full generation
- —Smart regeneration mode: skip trivial diffs (whitespace, comments, imports)
- —Regeneration mode selector in Settings (always / smart / semantic)
- —Processing chain now survives Gemini 429s — after() scheduled even on rate limit
- —Last-activity counter shows seconds within the first 3 minutes (1m 23s ago)
- —Jobs page: info modal adapted to configured AI provider
- —Mistral AI provider — Codestral model, free tier, OpenAI-compatible API
2026-07-11
- —Stall detection at 2 min with auto-restart while jobs page is open
- —Chain error logging via processing_health table — visible as banner in jobs page
- —Delete button extended to pending jobs
- —Gemini rate limit fix: batchSize 1, minDelay 4s
- —llms.txt, llms-full.txt, sitemap.xml, robots.txt, security.txt, ai-plugin.json
- —JSON-LD SoftwareApplication structured data on homepage
2026-06-30
- —Gemini (Google AI Studio) support — 1M context window, free tier
- —Smart regeneration Level 1 groundwork in processJob()
- —Rate limit profiles per provider (batchSize, minDelay, retryDelay)
- —Jobs page: progress bar, ETA, per-job retry and delete
2026-06-15
- —Bitbucket integration — OAuth login, webhook HMAC, repo listing
- —VCS abstraction layer (VCSClient interface, GitHub + Bitbucket adapters)
- —Pause / resume per repo
- —Exclusion list with glob patterns