Scan websites from your terminal — fast, repeatable, and AI-aware

Request scans, capture session videos, and get AI-summarized results. Use the lightweight CLI locally or call the API from your CI/CD pipelines.

Get the CLI

Blazing Fast

Optimized pipeline with staged progress and gentle updates.

Upload your project for AI analysis

Use the CLI commands below to analyze code or request a server scan. The uploader respects a .subtoignore file (one pattern per line) and always ignores .env.

subto upload [dir] — run a local AI analysis on sampled files (does not upload to server).
subto scan upload [dir] — upload sampled file snippets to the server and request a scan; returns uploadId and scanId.

`.subtoignore` format: one pattern per line; lines starting with # are comments; patterns can be folder names (e.g. node_modules), simple globs like *.lock, or specific files. The uploader uses simple matching (exact, prefix, contains, or endsWith).

To enable richer AI responses set OPENAI_API_KEY or OPENROUTER_API_KEY in your environment. You can choose a model with AI_MODEL (for example: openai/gpt-oss-120b:free).

Example:

subto upload . --max-files 200 --max-bytes 5242880

Server endpoints

Interactive links printed by the CLI point to two server endpoints:

/video/:scanId — redirects to a signed GCS URL for the session video (short-lived signed URL, typically ~1 hour).
/aichat/:scanId — opens the stored scan UI at /scanned/:scanId where the AI assistant UI is available for interactive Q&A about the scan.

Make sure your environment sets OPENROUTER_API_KEY or OPENAI_API_KEY if you want server-side AI features. The CLI and server will use AI_MODEL when present to select a model.

AI Summaries

Ask the assistant about issues, performance, or the session video.

Session Video

Record user flows for debugging and demonstrations.