What it does
Its ACT label is a recommendation, not permission to act. An optional gateway can steer the assistant's tool choices; file evidence goes to hosted inference, and short excerpts can miss context.
Agent tooling
jevkit, bragamat's Go toolkit
Give a coding assistant tools to find relevant source lines, check a claim against a file, rank options and sort records with Jev.
Bookmark: jevkit, bragamat's Go toolkit
Keep this in your collection.
Leave a noteWhat would you try with this?
: jevkit, bragamat's Go toolkit
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
Give your coding assistant a way to read less of a file. Begin with a harmless document and a question whose answer you know. The tool suggests lines to read or checks whether passages support a claim. Open the nearby text yourself: shortened lines can leave out something important.
Judgments need a key connecting the tool to TypeSafe, an outside AI service. It receives the selected file text. Keep permission to act in your own software: the tool’s ACT label is only advice. Its optional connection to a coding assistant continues without that advice when Jev fails.