Open chrome://extensions, enable Developer mode, click Load unpacked, and select the jeradin-extension folder.
Extension API
Endpoint
POST https://jeradin.com/api/public/extension/analyze — accepts a base64 PNG and returns a structured diagnosis + fix report.
Authentication
Header Authorization: Bearer <connection_token>. Tokens are per-account and can be revoked on the extension page anytime.
Response shape
Returns { tier, analysis, fix, latencyMs }. tier is "instant" or "smart" depending on which model handled the request.
MCP Connector
Endpoint
Jeradin exposes an MCP server at https://jeradin.com/mcp. Point any MCP-compatible client at that URL.
Permissions
MCP connections are read-only and scoped to your workspace. Jeradin never gets write access to your repo.
Available tools
list-notifications, get-profile, get-active-repo, set-active-repo. More coming.
Engine
Two-tier routing
Every capture starts in Instant mode. If the model self-reports low confidence or high complexity, Jeradin transparently escalates to Smart mode. You always get the best available answer.
Screen understanding
OCR + DOM accessibility tree parsing feeds a vision-language model. Jeradin reasons about the whole frame, not just the copied text.
Privacy
Frames are sent to Jeradin only when you press capture. They are never used for training and are discarded after the report is returned.