#580175 lnd: 0.21.4-beta -> 0.21.5-beta
#580174 agentsploit: 1.6.3 -> 1.7.0
#580171 ocamlPackages.jose: 0.10.0 → 0.11.0
#580169 ghgrab: 2.2.0 -> 2.3.0
#580165 maintainers: update baksa
#580156 topcoat-cli: 0.10.0 -> 0.11.0
#580155 multica-cli: 0.6.1 -> 0.7.1

Pulled the tool definitions from 28 of the most popular MCP servers and counted what they actually give an agent. 1,384 tools.

14 of the 28 hand the agent something that reads untrusted input AND something that can do real damage, in the same context.

github.com/agentsploit/agentsploit

1.384 herramientas MCP analizadas: así está el riesgo real

59 herramientas MCP leen y ejecutan a la vez. Mirá qué dice el analisis de seguridad en servidores MCP sobre 1.384 tools de 28 servidores antes de conec...

#mcp #seguridadenservidoresmcp #agentsploit #toolpoisoning #agentesdeia

AgentSploit: open-source offensive security framework for AI agents and MCP servers.
Prompt injection × permission graph × memory poisoning × live engagement dashboard.
pip install agentsploit
github.com/agentsploit/...
Apache 2.0

📦 New in #nixpkgs (2026-08-06):

• kubexhunt (#549651)
• openhack (#549666)
• agentsploit (#549659)
• {python3Package,}padne (#549111)
• rush-lyrics (#531990)
• esphome-device-builder (#535285)
• tmaze (#497038)
• gz-math (#401753)
• python313Packages.colcon-ed (#411846)


Full list on GitHub