Connect free-tier model providers through one OpenAI-compatible API. It routes requests between the providers you configure and switches when one hits a limit.
Useful for trying different models without wiring up every provider separately.
Bring your own provider keys. Free-tier quotas still apply; the project is intended for personal experimentation.
AI agents investigate your application for security vulnerabilities, validate findings and provide remediation guidance. It can work with code, web apps and APIs.
Useful for checking your own app before shipping a change.
Local runs require Docker and an LLM provider key. Model usage may cost money.
Captures observations from your agent’s work, summarizes them and brings relevant context into later sessions. Project history becomes something the agent can search.
Useful when you keep explaining the same decisions in a new chat.
Memory processing depends on your chosen provider or plan. See the README for local and hosted options.