#ai-safety

AI Code Exposure Monitor

ConsultantBPMhumansoftware

Live % of your code seen by AI assistants. Detects secrets, passwords, PII exposed to AI tools (Copilot, Claude, Cursor, etc.). 100% local. Works in VS Code, Cursor, Windsurf, VSCodium. MCP + team-work exposure features ready — just ask.

Glyphmap AI Guardrails

glyphmap-vscode

Stop AI coding agents from hallucinating files, drifting off-task, and making unsafe project changes.

SigmaShake SSG — AI Governance

sigmashake

AI governance, guardrails, and audit for Claude Code, Cursor, Copilot, Gemini, and any MCP-compatible assistant. Approve or deny tool calls in real time. .rules language support included.

Siphrix - AI Action Firewall

siphrix

Guard Claude Code with the Siphrix AI action firewall. One click installs the engine and puts every agent tool call through your policy before it runs.

CleanMyPrompt — AI Security & Token Intelligence

cleanmyprompt

Stop leaking secrets. Stop burning tokens. Intercepts API keys, PII & credentials before they reach Copilot, and compresses code with LSP-aware squeeze to cut token costs by 30–70%. 100% local.

MCPShield - MCP Security Scanner

MCPShield

Scan MCP servers for security vulnerabilities before connecting. Two-pass analysis: regex + LLM semantic detection.

AgentLens — See What AI Agents Really Do

sumonbis

A lens into AI coding agents. Click through what an agent actually did — under the hood — spot the unsafe steps, and learn to calibrate your trust. Explore sample runs or import a real one. For CS and non-CS students alike.