Developer Tools●●●Banger
Limits – Control layer for AI agents that take real actions
Wire-protocol interception means zero code changes; solves LLM control drift in production.
Solve My ProblemSlick
thesvp
923mo ago
Back to browse
GitHub Repository
🔪 Open-source safety firewall for AI agents. Intercepts tool calls before they execute, enforces YAML policies, and kills dangerous operations in real-time. Works with OpenAI, Anthropic, LangChain, and MCP. She doesn't guard. She kills.
10 starsPython
AvaKill – Deterministic safety firewall for AI agents (<1ms, no ML)
by duroapp·Mar 8, 2026·3 points·3 comments
AI Analysis
●●●BangerSolve My ProblemBig BrainShip It
Deterministic <1ms policy kill switch for AI agent tool calls, zero ML.
Strengths
- •Deterministic rule engine with red-team validation (63/63 blocked) closes a critical safety gap that existing solutions (SandboxAI, Courtyard) don't address as comprehensively
- •YAML policy format is operator-friendly; 81 rules across 14 categories provides immediate coverage without custom logic
- •Three independent enforcement paths (hooks, MCP proxy, OS sandbox) means works with existing agents (Claude Code, Cursor, Windsurf, Claude Desktop)
Weaknesses
- •Depends on tool-call intercepts; agents using direct file/network access bypass entirely
- •AGPL license limits commercial adoption
Category
Target Audience
Enterprises deploying AI agents, security teams enforcing tool-use boundaries
Similar To
SandboxAI · Courtyard · Anthropic token control
Similar Projects
Security●●●Banger
Rampart – Runtime firewall for Claude Code and AI agents in YOLO mode
Prevents `rm -rf ~` from your hallucinating agent in two commands, works with any CLI tool.
Solve My ProblemZero to One
trevxr
204mo ago
Security●●●Banger
Arbiter – Deterministic Guardrails for Agents
OPA policies plus signed tokens beat prompt engineering for agent safety.
Big BrainSolve My ProblemShip It
randromeda
112mo ago
Security●●Solid
Stop your agent running Terraform destroy or Git push origin main
Blocks terraform destroy and git push before agents execute destructive commands.
Solve My ProblemBig Brain
prudhvinomos
302mo ago
Developer Tools●●●Banger
CSL MCP Server – Write and Verify AI Safety Policies from Claude/Cursor
Mathematically verified policies enforced outside the model—formal proof replaces prompt engineering.
Big BrainZero to One
aytuakarlar
113mo ago
Security●●●Banger
SentinelGate – Universal Firewall for AI Agents (Open Source, Go)
Addresses real risk: AI agents currently run unrestricted—SentinelGate proxies all actions.
Solve My ProblemBig Brain
Sentinel-gate
213mo ago