Audits any skill before it installs. Trigger on any request to install, add, download or try an agent skill, on a pasted skill repo URL, on /bouncer, or on "is this skill safe". Searches for better alternatives first, then gates footprint, overlap, hidden…
Jack's mega design skill. One dispatcher that gives the agent taste — dark premium aesthetic, editorial typography, schematic illustrations — for any visual output: HTML pages, presentations, dashboards, course chapters, reports. Trigger on any design/build…
Route every task to the cheapest model that can actually do it. Benchmark-aware routing across frontier APIs, cheap CLIs and local models, with a cost receipt per session. Trigger on "route this", "cheapest model for", "save tokens on this", batch/mechanical…
any input (code, docs, papers, images) - knowledge graph - clustered communities - HTML + JSON + audit report
Move the agent's brain to a server that never sleeps. One command deploys Hermes Agent + Claude OS onto a fresh Linux VPS (hosting.com), hardens it, wires systemd + watchdog + nightly Dream, connects Telegram, and verifies end to end. Trigger on "deploy my…
Token-saving model router. Detects which AI providers and CLIs you have (ChatGPT/Codex, Gemini, Grok, OpenRouter, Hermes, local Ollama), then routes each task to the cheapest capable one — delegating grunt work to a cheaper CLI/local model, or switching the…
Overnight research runs that end in a morning brief. Watches competitors, comment demand, GitHub trends and niche signals while the user sleeps, writes one skimmable brief with receipts. Trigger on "run the scout", "what did the scout find", "/scout", or as…
The skill that builds skills. Watches for workflows the user repeats manually, drafts a new skill for the pattern, tests it, and submits it to the Bouncer — nothing installs without passing the gates. Trigger on "make this a skill", "I keep doing this…