whereismyai
Finbot Ctf
Finbot Ctf is an intentionally vulnerable agentic AI platform and optional Tenuo defended mode fork for learning and testing AI security.
FinBot Ctf is an intentionally vulnerable agentic AI platform created by OWASP GenAI for learning, testing, and practicing AI security. This fork adds an optional defended mode driven by Tenuo warrants, which are signed lists of allowed tools and argument values built from database records rather than prompts. Every tool call is checked against the active task warrant before running, limiting what fooled agents can actually do across workflows like invoice approvals, payments, and vendor chats. It supports local models via Ollama and OpenAI keys, running through Docker, Python, or a web UI. The platform helps developers and security researchers study vulnerabilities and test defenses using tools like Promptfoo.
Finbot Ctf at a glance
- Best for: Software engineer
- Runs on: Web, Docker, CLI, open source
- Pricing: Free and open source
- Released: October 2026
- Category: AI coding tool
- Website: github.com/tenuo-ai/finbot-ctf
Key features
- Optional Tenuo defended mode via environment variables
- Task-scoped signed warrants for agent tool calls
- Database-backed authority validation independent of prompts
- Delegation chain authority restriction for sub-agents
- Integration with OpenAI and local models via Ollama
- Promptfoo testing support across multiple entry points
- Docker and Redis support for local setups
Use it for
- Learning and testing agentic AI security vulnerabilities
- Running Promptfoo red-teaming attacks against AI agents
- Comparing agent security with and without Tenuo defense
- Executing demo workflows to verify tool call denials
Alternatives to Finbot Ctf
- Promptfoo — It simulates real users to uncover agent and RAG vulnerabilities.
- Daytona — It provisions sandboxes in under 90 milliseconds with full security isolation.
- PentAGI — It autonomously manages penetration testing workflows within a secure Docker sandbox.
- ARTEX — Runs autonomous multi-agent penetration testing with a dual-graph architecture.
- Ctxvitals — Measures exact token costs and security risks before agent launch.
- Cloudflare Agents — Runs agents on a global network with durable storage and real-time connections.
- AgentFS — Provides isolated filesystem access and instant snapshots via SQLite.
- Numbat — It enables local detection and blocking of AI agent actions on endpoints.
- CyberStrike — Autonomous AI agents execute recon, web, network, and cloud penetration testing.
- Ctxfw — Reduces AI context token mass while preserving code structure and dependencies.
- PentestGPT — Runs autonomous multi-stage pipelines or interactive multi-LLM workflows for security tests.
- Kungfu — Maintains work state and context when switching between different AI agents.
All Finbot Ctf alternatives