Observed arrival · 2026-10-24
moatleak: a map of your AI prompt trail
This open-source tool scans local Hermes and Claude Code logs, then reports which companies and parent groups processed prompts and where those companies are headquartered.
- For
- Developers auditing their local AI-tool usage
- Worth noticing
- Its sample flags one conversation routed through three companies; country labels track corporate headquarters, not necessarily the servers handling prompts.
Field notes
Installation is via pipx or pip and requires Python 3.11+ on Linux or macOS; the scanner reads Hermes agent logs and state.db, plus Claude Code transcript files, then writes HTML, CSV, SQLite, and JSON outputs. The sample uses the maintainer’s own logs from 22 Sep–7 Oct 2026 and lists 10 companies across two countries, but those country labels refer to company and parent-group headquarters, not necessarily server locations. The site says scanning is read-only and makes no network calls.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue