arkheia_verifyLABSCO SUMMARY
A risk score an agent can branch on, not a second opinion to read.
The value is that it returns a level rather than a paragraph, which is what makes it usable inside a loop: LOW passes through, MEDIUM gets a caveat, HIGH blocks the merge from a sub-agent until someone looks. Try it on a plausible-sounding API that does not exist — that is the failure mode it is built for, and the one review by eye is worst at catching.
What it is
A screening service for model output. You hand it the model name, the prompt and the response, and it returns a fabrication risk score derived from behavioural fingerprinting rather than a fact-check lookup. It covers Claude, GPT, Gemini, Grok, Llama, Mistral and others.
What you get
arkheia_verifyscores a response for fabrication riskarkheia_audit_logreviews your detection historyrun_grok,run_geminiandrun_togethercall the model and screen the answer in the same steprun_ollamadoes the same against a local model- A three-level result — LOW, MEDIUM, HIGH — that an agent policy can branch on directly
- A ready-made verification protocol for
CLAUDE.mdin the README, including how to handle sub-agent output
Requirements
Node.js 18+, Python 3.10+ and git. npx @arkheia/mcp-server sets up the Python environment and configuration; the client then runs python -m mcp_server.server from ~/.arkheia/mcp with ARKHEIA_API_KEY set. Keys are free and provisioned by one HTTP call against your email address, and shown once. The free tier covers 1,500 detections a month with 5 concurrent, no card. Screening means the prompt and the response leave your machine. Version 1.3.0.
Setup effort
Follow the published steps — npx @arkheia/mcp-server
USE CASES
01
Checking an answer before you act on it
Score the response and treat a high result as a reason to investigate the specific claims.
02
Gating what background agents hand back
Verify each agent's output on its own, and hold the merge when one comes back high.
03
Comparing how different models behave on the same question
Call the model and screen the answer in one step, across hosted and local models alike.
6 tools
Listed in project docs. Your client may show a different list.
arkheia_audit_logrun_grokrun_geminirun_ollamarun_togetherBEFORE YOU INSTALL
What to know first
Tools you get
6 tools named in the project's documentation.
Requires
- Node.js 18+
- Python 3.10+
- ARKHEIA_API_KEY in your environment
Optional
ARKHEIA_PROXY_AUTH_TOKEN in your environment
Where it runs
Both documented clients run the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
npx @arkheia/mcp-serverGet a free API key — save it, it is shown only once:
TERMINAL
curl -X POST https://arkheia-proxy-production.up.railway.app/v1/provision -H "Content-Type: application/json" -d '{"email": "<YOUR_EMAIL>"}'MCP Clients
Claude Code, Claude Desktop
JSON
{
"mcpServers": {
"arkheia": {
"command": "python",
"args": [
"-m",
"mcp_server.server"
],
"cwd": "~/.arkheia/mcp",
"env": {
"PYTHONPATH": "~/.arkheia/mcp",
"ARKHEIA_API_KEY": "<YOUR_ARKHEIA_API_KEY>"
}
}
}
}Clients start this without a shell, so ~ is not expanded: write cwd and PYTHONPATH as the full path of the folder the installer created (for example /Users/you/.arkheia/mcp; on Windows the README gives C:/Users/YourName/.arkheia/mcp). PYTHONPATH is what lets python find mcp_server.server, so it has to be the full path even in a client that ignores cwd.
The configuration on this page does not set ARKHEIA_PROXY_AUTH_TOKEN.
