Skrypt generates docs from your code and runs every snippet to prove it works. When your API changes, the docs update themselves.
$ npx skrypt-ai test
Scanning docs in ./docs...
Found 42 code examples across 18 pages
Testing against node-20, python-3.12...
✗ docs/quickstart.mdx:24 TypeError: fetch is not defined
✗ docs/auth.mdx:67 Property 'token' missing in type
✗ docs/api/users.mdx:31 Expected 2 arguments, got 1
39 passed 3 broken
Run `npx skrypt-ai generate --verify` to fix automatically.
How it works
One CLI does all three.
Run skrypt test on any repo. It catches broken code examples: wrong arguments, missing imports, stale API calls. Free and unlimited.
Point it at your codebase. Skrypt parses functions, classes, and types across 10 languages, writes the docs, and verifies every code example compiles and runs.
Your API changes. A code example breaks. Skrypt catches the error, regenerates the snippet, and keeps going until it passes. No tickets, no manual fixes.
Features
One CLI. Every snippet tested before it ships.
Snippets are compiled and executed for real. Not linted, run. If it doesn't pass, it doesn't ship.
API changed? Skrypt catches the breakage, feeds the error to AI, and regenerates until the snippet works again. You don't touch it.
Point at your codebase, get docs in minutes. Works across TS, Python, Go, Rust, Java, Kotlin, Swift, PHP, C#, and Ruby.
AST-level parsing for TypeScript, Python, Go, Rust, Java, Kotlin, Swift, PHP, C#, Ruby. Pulls out functions, classes, types, and signatures.
Migrate from Mintlify, Docusaurus, GitBook, ReadMe, or Notion. One command. Maps every component automatically.
OpenAI, Anthropic, DeepSeek, or run Ollama locally. Your key, your calls, your machine.
Runs each snippet against Node 18/20/22, Python 3.10-3.12, Bun, and Deno. Like a CI matrix, but for your docs.
Compares your code to your docs and tells you what's missing. Coverage gaps, undocumented functions, stale signatures, orphaned pages.
Only regenerates docs affected by recent code changes. Saves 95% of LLM cost on incremental updates.
Stemming, fuzzy matching, and field boosting built in. Handles typos and partial queries without extra config.
Sandpack-powered examples your users can edit and run in the browser. Supports Python, Go, and JS.
Ships llms.txt, an MCP server, JSON-LD, and Open Graph on every page. So Cursor, Claude, and Google all read your docs correctly.
Compare
What ships free vs what others skip.
Security
Nothing sensitive leaves your machine. Your keys, your infra, your calls.
When you bring your own API key, LLM calls go directly to your provider. Your key and your code never touch Skrypt servers.
API keys are stored in your OS keychain (macOS Keychain, Windows Credential Manager, Linux Secret Service). Never in plaintext files.
API keys are hashed with SHA-256 for verification and encrypted at rest with AES-256-CBC. Keys are never stored in plaintext, on the server or on disk.
Generated docs are a standard Next.js site deployed to your own infrastructure. Run skrypt security to audit your setup at any time.
Pricing
Find every broken code example for free. Fix them automatically with Pro.
Free
$0
Find every broken code example. Free forever.
skrypt test, unlimited, any repoPro
$199/mo
Docs that fix themselves when your API changes.
Enterprise
Custom
We maintain your docs so you don't have to.
Run the test for free. Fix them automatically with generate.
$ npx skrypt-ai test
$ npx skrypt-ai generate --verify