Features
Every capability below is shipped and running on the platform's own workstation dashboard today. Where a capability isn't built yet, it's named as a roadmap item on the Documentation page instead of appearing here as if it were live.
Ctrl+K opens a unified, fuzzy-filtered search across nav modules, quick actions, real n8n workflows, real Ollama models, live Docker containers, business units, and full-text repo docs — one search box, not one per tool.
Real n8n executions, git commits, Docker events, MCP transitions, and AI chat turns merged into one newest-first feed — nothing on a timer just to look alive.
Every notification is derived from a detected event — workflow failures, container state changes, model load/unload, new commits, resource warnings crossing 90%.
Real CPU/RAM sampling, GPU/VRAM via nvidia-smi, disk via Get-CimInstance, the full Docker container list, and n8n's active-execution count — each with a rolling 24-point trend, not just a raw number.
Pull, delete, benchmark, and test-prompt directly against Ollama, with last-used tracking. No invented benchmark numbers — every figure comes from a real run.
Success rate, run counts, and trigger type per n8n workflow; containers/images/volumes/networks with confirmation-gated restart/stop; real git branch, commits, issues, and PRs.
A dockable chat panel runs a real request/response loop against local Ollama. Three one-click Copilot actions — summarize system health, explain recent workflow failures, search docs — are built from real data, not scripted responses.
Claude and OpenAI appear as visibly disabled “coming soon” options rather than a faked integration.
Per-business Research, Scheduler, and Publisher agents: Ollama-assisted content ideas held for manual review, auto-scheduling for records with no date, and confirmation-gated publishing.
Manifest-driven: onboarding a structurally new business is a configuration entry and a data folder, not new framework code.