From pipecrew
Launches a machine-wide dashboard showing all active Claude Code sessions with token usage, sub-agents, and needs-approval badges. Complements per-run /site-view.
How this skill is triggered — by the user, by Claude, or both
Slash command
/pipecrew:siteview-fleetThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
The **fleet view** — one dashboard for *every* Claude Code session running on your
The fleet view — one dashboard for every Claude Code session running on your
machine, not just the current /deliver run. Complements the per-run
/site-view:
/site-view → one /deliver run in depth (6 stages, crew, pyramid, gates)./siteview-fleet → all sessions at a glance (tokens, sub-agents, "needs
you" badges), click any card for its agent-dispatch tree + activity timeline.PipeCrew /deliver sessions show up here with the same pharaoh crew icons as
/site-view (bruno, crit, archie, …); plain Claude Code sessions get a distinct
desert-expedition cast.
This is powered by the standalone pipecrew-siteview tool (a separate npm
package), so it also works outside PipeCrew. The skill just locates your install
and starts it.
/pipecrew:siteview-fleet [--port=<n>] [--projects-dir=<path>] [--demo]
Flags (passed straight through to pipecrew-siteview):
--port=<n> — starting port (default 5174; auto-increments if busy).--projects-dir=<path> — Claude projects root (default ~/.claude/projects).--demo — serve a fabricated, self-animating fleet (no real sessions, zero tokens).When the user invokes /siteview-fleet:
Launch via Bash with run_in_background: true (the server keeps serving):
node {plugin_dir}/skills/siteview-fleet/launch.js [--port={n}] [--projects-dir={p}] [--demo]
launch.js resolves pipecrew-siteview in this order and hands off to the
first real install: $PIPECREW_SITEVIEW_DIR → ~/pipecrew-siteview clone →
a global install on PATH. (It does not silently install or npx-fetch.)
Report the URL it prints (default http://127.0.0.1:5174/). The tool
auto-opens the browser.
If it exits non-zero ("not installed"), offer the opt-in install: ask the
user "Install the fleet view now with npm install -g pipecrew-siteview?"
--install (it runs the global
install, then launches):
node {plugin_dir}/skills/siteview-fleet/launch.js --install [--port={n}] [--projects-dir={p}] [--demo]
npm install -g pipecrew-siteview,
clone to ~/pipecrew-siteview, or set $PIPECREW_SITEVIEW_DIR.Do NOT pass --install on the first attempt — only after the user opts in.
The fleet view can badge a session (and pop an OS notification) the moment it's waiting on you. That's driven by a Claude Code hook the standalone installs:
pipecrew-siteview --install-hooks # idempotent; merges alongside PipeCrew's own hooks
Restart Claude Code afterward so the hook loads. See the pipecrew-siteview
README for the full notification / sound options.
/siteview-fleet does not replace /site-view, /siteview-list, or
/siteview-cleanup — those manage per-run PipeCrew servers driven by
checkpoints.jsonl + the scratchpad. The fleet view is a higher, machine-wide
layer that reads Claude Code session transcripts directly and needs no run dir.
npx claudepluginhub pipecrew-ai/pipecrew --plugin pipecrewPresents a read-only dashboard of all Claude sessions (local and remote) showing token types, CRS relay health, account utilization, and per-session details. Useful for fleet monitoring.
Starts, stops, or checks dot-dash server for real-time browser dashboard monitoring Claude Code sessions, streaming transcripts via WebSocket, and prompt injection.
Live browser UI for PipeCrew agent pipeline runs. Shows characters as queued/working/done with pyramid tiers, fed via SSE. Auto-started by /deliver skill.