romanklis/openclaw-contained
TaskForge runs AI agents in sandboxed Docker containers with capability-based security. Agents start with minimal permissions and must request new capabilities (packages, network access, tools) through a human-in-the-loop approval process. Every approval triggers a container image rebuild, and every LLM interaction is logged for audit.
TaskForge helps organizations safely use AI agents to automate complex tasks, from data analysis to application deployment. It takes a high-level request (like "build a Bitcoin tracker") and produces a working application or report, ensuring that every step taken by the AI agent is logged and approved by a human. Security teams, operations managers, and compliance officers would use this to supervise AI-driven workflows.
Use this if you need to run AI agents that can build applications or execute code, but require strict oversight, audit trails, and robust security to prevent unauthorized actions.
Not ideal if you need to run simple, low-risk AI tasks where human approval for every new capability or a detailed audit trail is overkill.
Stars
24
Forks
7
Language
Python
License
—
Category
Last pushed
Mar 24, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/agents/romanklis/openclaw-contained"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
heshengtao/super-agent-party
⭐ All-in-one AI companion! Super Agent Party = Self hosted neuro sama + openclaw! ⭐...
darfaz/clawmoat
🦀 Security moat for AI agents. Runtime protection against prompt injection, tool misuse, and...
dataelement/Clawith
OpenClaw for Teams
scottgl9/LeanClaw
LeanClaw is a high-efficiency, security-first AI assistant runtime built for fast local...
quoroom-ai/room
Open-source earning-focused swarm intelligence engine. Self-governing AI collectives (queen,...