Humane-Proxy MCP
An AI safety middleware that intercepts LLM prompts to detect self-harm and criminal intent. It features a 3-stage safety pipeline and provides an MCP server with tools like check_message_safety, get_session_risk, and list_recent_escalations to protect users before messages reach the upstream LLM.
Package and access
Local package with source-attributed installation metadata. Koskamo does not execute packages during review.
Checks and signals
Each line is separate evidence, not a trust score.
- Popularity 28 GitHub stars
What you can do with it
- Detect self-harm ideation in user prompts
- Identify criminal intent in LLM interactions
- Block unsafe messages and provide care responses
- Alert operators to high-risk sessions
Source and maintenance
- Last repository push
- Latest release
- v0.6.0
- Maintenance checked
Packages and installation
Source-provided package data. Koskamo has not executed or live-probed this package.
humane-proxy 0.6.0
{
"version": "0.6.0",
"transport": "stdio",
"identifier": "humane-proxy",
"registryType": "pypi",
"packageArguments": [
{
"type": "positional",
"value": "mcp-serve",
"valueHint": "command"
}
],
"environmentVariables": [
{
"name": "OPENAI_API_KEY",
"description": "OpenAI API key for Stage 3 reasoning (optional)"
},
{
"name": "GROQ_API_KEY",
"description": "Groq API key for LlamaGuard Stage 3 (optional)"
}
]
}Tools
Reported by the attributed source; not independently observed by Koskamo.
Maintenance history
Public check and source-refresh events. Editorial and ownership records remain private.
- · repository · passed · metadata changed · updated
- · backfill · recorded · Pre-expansion baseline; unobserved rich metadata remains unknown
Connect from your client
Each snippet configures the same source-provided package; replace placeholders before use.
Cursor
{
"mcpServers": {
"humane-proxy-mcp": {
"command": "uvx",
"args": [
"humane-proxy==0.6.0",
"mcp-serve"
],
"env": {
"OPENAI_API_KEY": "<YOUR_OPENAI_API_KEY>",
"GROQ_API_KEY": "<YOUR_GROQ_API_KEY>"
}
}
}
}Cline
{
"mcpServers": {
"humane-proxy-mcp": {
"command": "uvx",
"args": [
"humane-proxy==0.6.0",
"mcp-serve"
],
"env": {
"OPENAI_API_KEY": "<YOUR_OPENAI_API_KEY>",
"GROQ_API_KEY": "<YOUR_GROQ_API_KEY>"
}
}
}
}VS Code
{
"servers": {
"humane-proxy-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"humane-proxy==0.6.0",
"mcp-serve"
],
"env": {
"OPENAI_API_KEY": "<YOUR_OPENAI_API_KEY>",
"GROQ_API_KEY": "<YOUR_GROQ_API_KEY>"
}
}
}
}Claude Desktop
{
"mcpServers": {
"humane-proxy-mcp": {
"command": "uvx",
"args": [
"humane-proxy==0.6.0",
"mcp-serve"
],
"env": {
"OPENAI_API_KEY": "<YOUR_OPENAI_API_KEY>",
"GROQ_API_KEY": "<YOUR_GROQ_API_KEY>"
}
}
}
}Let your agent find its tools
Find servers like Humane-Proxy MCP from your agent. Add Koskamo once to search the index for endpoints, access requirements, and client setup. Public discovery needs no key.