bug-bounty517
xss282
rce126
bragging-post117
account-takeover104
open-source95
google85
authentication-bypass85
csrf85
facebook75
privilege-escalation74
exploit74
stored-xss74
web-security68
access-control67
ai-agents65
reflected-xss63
writeup61
microsoft59
input-validation53
ssrf52
sql-injection48
smart-contract48
defi48
tool47
cross-site-scripting47
information-disclosure45
api-security45
ethereum45
privacy44
cve43
web-application38
llm38
automation37
malware36
burp-suite36
vulnerability-disclosure36
opinion35
web334
html-injection33
smart-contract-vulnerability33
responsible-disclosure33
infrastructure32
machine-learning32
code-generation32
lfi31
apple31
authentication31
phishing31
waf-bypass31
0
5/10
Agent Safehouse is a macOS-native sandboxing tool that enforces kernel-level deny-first access controls to restrict LLM agents (Claude, Codex, Gemini, etc.) to specific project directories, preventing accidental or malicious access to sensitive files like SSH keys and AWS credentials outside the target workspace.
sandboxing
macos
local-agents
llm-security
access-control
kernel-enforcement
privilege-restriction
ai-agent-safety
tooling
Agent Safehouse
Claude
Codex
Gemini
Copilot
Cursor
Cline
Aider
eugene1g