Autonomous coding agent
▸ Talk to it. It explores, fixes, and tests on its own.
Wisp takes a natural-language task and works autonomously — exploring the repo, editing code, running tests, and producing a final answer or PR. Pick a sample below to watch each stage: reasoning, tool call, result, final answer. In production, every LLM and tool call is recorded by Argosvix (this demo replays a recorded run).
Sample instructions