Self-hosted multi-agent system that autonomously plans and runs penetration tests using tools like nmap, Metasploit and sqlmap in Docker.
Claude Code and other coding agent skill plus proxy that makes the agent answer in terse caveman-style prose to cut token use.
Runs open-weight LLMs such as Gemma, Qwen, DeepSeek, and gpt-oss locally or in the cloud, with a CLI, REST API, and Python and JavaScript libraries.
Persistent memory system for AI coding agents, shipped as a single Go binary backed by SQLite with FTS5 search.
Open-source local AI engine that runs LLMs, vision, voice, image, and video models behind OpenAI-compatible APIs on CPU or GPU.
Scans git repos, files, and stdin for hardcoded secrets such as passwords, API keys, and tokens.
Self-hosted AI gateway that exposes one API to clients while routing requests to multiple upstream model providers.
Autonomous API and web-app penetration testing tool that runs a swarm of AI agents to chain exploits and prove vulnerabilities.
Scrapes Google Maps listings for business name, address, phone, website, rating, reviews, coordinates, and emails via CLI, Web UI, or REST API.
Go proxy that converts Google AI Studio web access into OpenAI, Anthropic, and Gemini compatible APIs with multi-account support.
Distributed security asset discovery platform covering subdomain enumeration, port scanning, vulnerability scanning, and page monitoring.
Terraform core: a CLI and graph engine for declaring, planning, and applying infrastructure changes from versioned configuration files.
Defines AI-powered GitHub Actions automation in Markdown with YAML frontmatter, compiled by the gh CLI extension into standard workflows.