Kritt-ai/open-kritt
open·kritt packages AI-agent security research into ranked, de-duplicated vulnerability findings.
The project runs focused prompts across parallel agents, then validates, enriches, and exports the resulting findings. It supports local or remote repository scans with Codex or Claude Code, plus model access through OpenAI, Anthropic, OpenRouter, xAI, or a Codex login. The README stresses that the stack has no built-in application authentication and should be kept private. Tool-enabled agents run as root in disposable containers with internet access, so the maintainers recommend a dedicated Docker host or VM. GitHub · Agents repos' note.
The project runs focused prompts across parallel agents, then validates, enriches, and exports the resulting findings. It supports local or remote repository scans with Codex or Claude Code, plus model access through OpenAI, Anthropic, OpenRouter, xAI, or a Codex login. The README stresses that the stack has no built-in application authentication and should be kept private. Tool-enabled agents run as root in disposable containers with internet access, so the maintainers recommend a dedicated Docker host or VM. GitHub · Agents repos' note.
score 4