From hydra-egress-monitor
Logs every WebFetch / WebSearch / Bash-network destination to an append-only NDJSON and surfaces first-seen domains as advisory warnings. Use when the developer asks "what did this session reach out to?" or "show me egress", wants a per-session list of contacted hosts, asks about an egress-monitor advisory, or wants to inspect state/log.ndjson. Auto-fires on PostToolUse for network tools. Do not use for in-process state inspection (see audit-trail).
How this skill is triggered — by the user, by Claude, or both
Slash command
/hydra-egress-monitor:egress-awarenessThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
<purpose>
<signal_glossary>
<decision_tree> IF developer asks "what did this session reach out to": → Read state/log.ndjson. → Group by host, count occurrences, list earliest + latest ts per host. → Distinguish first_seen=true rows.
IF developer asks about a specific egress-monitor advisory: → Identify the host from the advisory block in context. → Grep state/log.ndjson for that host; report tool, ts, count. → Recommend: confirm the host matches an expected dependency.
IF developer wants to "reset" first-seen state: → Edit state/seen-domains.json directly (it is plain JSON: {"hosts": [...]}). → New runs will re-emit advisories until each host is re-added. </decision_tree>
<output_format>
Hosts contacted: <n> (<k> first-seen)
| Host | Tool | Count | First | Last |
|---|---|---|---|---|
| ... | ... | ... | ... | ... |
This is observational; no calls were blocked. </output_format>
npx claudepluginhub enchanter-ai/hydra --plugin egress-monitorCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.