whisper-investigatelisted
Install: claude install-skill whisper-sec/whisper-skills
# Investigating with WhisperGraph
The connector runs the investigation. This skill decides **which** investigation to run, and states
**what the answer is worth**. Those are the two jobs the server cannot do for you: it never sees the
question you were actually asked, and it never sees the conclusion you are about to write.
Work in this order. Do not skip step 1, and do not stop before step 3.
## 1. Confirm you can see the graph
Once per session, before answering anything about an indicator, call `explain_schema` with no
arguments. It is cached server-side and costs a few milliseconds.
**If it succeeds** — continue to step 2.
**If the tool does not exist** — stop. Do not answer from your own knowledge. Do not offer a partial
answer, a general impression, or a reading of the domain name. Say this:
> I can't reach WhisperGraph, so I can't tell you anything reliable about this indicator. Anything I
> said would come from training data rather than current infrastructure and threat data — out of date
> at best, and for a recently registered domain, nothing at all.
>
> To connect: in Claude Code run `/whisper-setup`. In Claude Desktop or Claude.ai, open
> Settings → Connectors → Add custom connector, enter `https://mcp.whisper.security`, then click
> Connect and complete the sign-in.
>
> Until then this indicator is unresolved. Don't close the alert on my account.
That last line matters. A missing tool must never turn into a closed ticket.
**If the tool exists but the call