From vs-token-safer
Opens/closes a local interactive dashboard showing token savings, language mix, per-tool savings, and a 3D include graph. Use when the user wants to visualize vs-token-safer data.
How this skill is triggered — by the user, by Claude, or both
Slash command
/vs-token-safer:vs-vizThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A local, interactive view of what vts knows + how much it saved. **Local-only and opt-in**: the server
vts serve)A local, interactive view of what vts knows + how much it saved. Local-only and opt-in: the server
binds 127.0.0.1, serves a fully self-contained page (inline CSS/JS + a locally-vendored Three.js —
no CDN, no external fetch), and runs ONLY while explicitly started — the always-on MCP server never starts
it. Same zero-transmission trust model as the rest of vts (localhost ≠ off-machine).
Under the hood (if running the CLI directly):
vts serve --open --projectPath <root> — start + launch browser. Run it with run_in_background: true
(it's a long-running server). --port N if 8731 is busy. Falls back to
node "$CLAUDE_PLUGIN_ROOT/server/cli.js" serve … when vts isn't on PATH.vts serve --stop — shut it down. Ctrl-C also works on the foreground process.~/.vs-token-safer/savings.json).vts warmup --projectPath <root> --backend <clangd|typescript|…> or just keep using the search tools.vts savings / vts_admin {op:"savings"}.VTS_VIZ_MAX_NODES (200) highest-fan-in nodes so the WebGL sim stays smooth.npx claudepluginhub jsungmin/vs-token-safer --plugin vs-token-saferProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.