Security●●Solid
HAL – Harmful Action Limiter: Lean command guard for AI coding agents.
Structural command parsing beats regex for catching dangerous agent actions.
Big BrainSolve My Problem
otherland26
393mo ago
Back to browse
GitHub Repository
AI agent safety shield — intercepts dangerous commands, requires password verification, sends alerts across all platforms and AI tools
1 starsShell
I built a safety shield for AI agents that intercepts dangerous commands
by Thomas_Zhou·Jul 6, 2026·1 point·0 comments
AI Analysis
●●SolidSolve My ProblemNiche Gem
Password-gated command interception beats silent AI agent failures.
Strengths
- •Multi-platform alert system covers Feishu, Telegram, Discord, WhatsApp, Signal, Slack
- •Tier-based blocking distinguishes filesystem destruction from git operations
- •Works at system level via shell script and git hooks, not just one AI tool
Weaknesses
- •Only 1 GitHub star suggests very early adoption and untested edge cases
- •No Windows support mentioned despite listing Windows destructive commands
Category
Target Audience
Developers using AI coding agents with shell access
Similar To
Cline · Cursor · Continue
Similar Projects
Security●●●Banger
AI Coding Agent Guardrails enforced at runtime
Blocks dangerous AI agent commands like rm -rf before execution in under 2ms.
Solve My ProblemSlick
cavalrytactics
502mo ago
Developer Tools●●●Banger
Kintsugi – a local-first safety net for AI agents and humans
Deterministic rule engine blocks rm -rf before it runs, not after. Works with every agent.
WizardrySolve My ProblemZero to One
arr0wassass1n
1020d ago
Security●●●Banger
OnGarde – Runtime content security proxy for self-hosted AI agents
Transparent proxy blocks agent leaks to LLM APIs without touching agent code.
Solve My ProblemSlick
antimaterial
214mo ago
Security●●Solid
Safety layer between AI agents and databases
AST analysis blocks injection attacks before they hit your production database.
Solve My ProblemShip It
burhanultayyab
392mo ago
Security●●Solid
Clawsec - Open-source plugin for OpenClaw that blocks dangerous actions
Every tool call is caught by middleware, scored against built-in rulesets like destructive-commands, secrets/* and exfiltration/* in under 5ms, then enforced as block/confirm/allow via a clawsec.yaml — neat, pragmatic attack surface reduction. The demo and auto-generated config make onboarding trivial, but it currently reads as an OpenClaw-first solution; broader agent-framework integrations or stronger isolation guarantees would make this a must-install.
Solve My ProblemNiche GemSlick
subho007
114mo ago