Audience
Engineering and security teams running AI agents in production: platform and AI engineers who build agents, application security and AppSec teams who have to govern them, and compliance teams who need evidence of what an agent was allowed to do.
About AgentFox
AgentFox is an open-source control plane for AI agent development and operations. It inventories agent repositories and maps agents, tools, MCP servers, skills, and handoffs without executing the code. For supported agent frameworks, it applies tool-access policies, validates tool arguments against configured limits, and can hold or reject actions that require human review. Runtime checks inspect model and tool traffic for prompt-injection content and possible secret or personal-data exposure. AgentFox can run in observation mode before enforcement is enabled, and records policy decisions in a hash-chained audit log. Developers can install it with Python or deploy the Docker Compose stack. It is intended for engineering and security teams that need visibility, controls, and reviewable records for AI-agent workflows. AgentFox is released under the Apache-2.0 license; a hosted preview is also available.
Pricing
Company Information
Videos and Screen Captures
AgentFox agent canvas mapping an LLM, tools, documents, and guardrails, with attack-test results.
