---
name: privi-secret-scan
description: Use before sharing logs, prompts or support transcripts that might contain secrets.
---

# Inspect text for likely credential leakage

Use the Privi Agent Survival remote MCP tool `privi_secret_scan` when this workflow is relevant. The server is a stateless calculation service. It does **not** perform external actions or manage the calling agent's credentials.

## Procedure

1. Confirm the input is user-authorized and contains no unnecessary secrets.
2. Connect an MCP-capable agent to `https://privi-ai-21-gateway.onrender.com/mcp` or use `POST /v1/tools/privi_secret_scan` with your own API key.
3. Provide the following JSON tool arguments:

```json
{"text":"Authorization: Bearer exampletokenvalue"}
```

4. Review the returned fields and error status; incorporate the results into your workflow without claiming an external action occurred.
5. Only returns finding categories and counts; manually review false negatives.

## Production cautions

All Privi tool executions require a prepaid API key; public discovery does not grant free execution. Stripe and Base checkout are disabled until verified. Never embed tokens, private keys or confidential customer records in a public repository. For high-impact decisions, confirm results with your own trusted data and human approvals.

© 2026 Privi World Corp. All rights reserved.