Scan, rate, and harden MCP servers for AI agent safety
-
Updated
Apr 20, 2026 - Python
#
tool-safety
Here are 4 public repositories matching this topic...
Keyless-by-default LLM security gateway + operator CLI to gate coding agents: inspect prompts/context/tool calls and enforce allow/review/block decisions.
policy-engine ai-security security-gateway guardrails llm-security prompt-injection-llm-security agent-security openclaw tool-safety
-
Updated
Mar 7, 2026 - TypeScript
MCPWarden is a policy-checkable workflow runtime and guardrail layer for MCP tools. It imports MCP servers, requires explicit trust and effect metadata, plans approvals, blocks unsafe workflows, redacts sensitive outputs, and provides a plug-and-play CLI/local service for agent tooling.
redaction typescript mcp policy-engine ai-agents prompt-injection workflow-runtime model-context-protocol agent-safety tool-safety
-
Updated
May 9, 2026 - TypeScript
Enforce tool-usage contracts on agent tool calls — block before side effects.
-
Updated
Mar 24, 2026 - Python
Improve this page
Add a description, image, and links to the tool-safety topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the tool-safety topic, visit your repo's landing page and select "manage topics."