Team & Operations skills for your AI agent.
Four skills addressing four different scales of the same complaint — that effort goes in and output doesn’t come out. High Output Management works at the scale of your own week: which hours multiply, how to match 1:1 cadence to how experienced someone is, why every decision routing through you leaves work waiting. Drive Motivation works at the scale of one person and the incentive quietly crowding out the motivation they arrived with. Team Topologies works at the scale of the boundaries between teams. Traction EOS works at the scale of the whole company — Rocks with a single owner, a weekly scorecard, a meeting that ends in decisions.
With these installed, an agent asked to “fix the team” stops proposing a reorg and better communication. It asks whether the problem is one manager’s leverage, one person’s autonomy, cognitive load spread across too many domains, or the absence of any operating rhythm — and it will tell you when a reorg is the wrong instrument for a motivation problem.
02 Choosing
Which team skill fits the problem I have?
Route by the size of the broken thing: your week, one person, the seams, the company.
The pairing worth having is Traction EOS with High Output Management. Hold Team Topologies until you have more than three teams — below that, redrawing boundaries is expensive theatre.
One gives the company a rhythm, the other gives you leverage inside it.
03 The four skills
All four Team & Operations skills.
Each installs on its own, in about thirty seconds, and works in any agent that reads the open agentskills.io format.
04 Used by
Where these four
get sequenced.
Guides and guided journeys that run Team & Operations skills alongside the rest of the library — in the order that stops you fixing the wrong thing first.
05 Questions
Before you install.
The four we get asked most about this stack. The generic ones — pricing, licence, agents — live on the library page.
I lead a team of four. Is Traction EOS overkill?
Usually, yes. EOS assumes a leadership layer with separate functional owners, quarterly Rocks and a scorecard someone maintains — machinery that costs more than it returns at four people, where everyone already knows what everyone is doing. High Output Management scales down much better: its ideas about leverage, delegation and task-relevant maturity apply to any manager with two reports. Revisit Traction EOS when you can no longer hold the whole company in one meeting.
Do Traction EOS and Team Topologies pull in different directions?
In one specific place, yes. EOS wants a clean accountability chart with one name over each function, which tends to organize people by discipline. Team Topologies argues that functional splits are exactly what stops anyone owning a change end to end, and wants stream-aligned teams that cross those lines. If you run both, let EOS supply the cadence and Team Topologies supply the boundaries, rather than letting the org chart decide both.
Can an AI agent be useful on people problems it can’t observe?
It can’t read the room, and neither framework pretends otherwise. What it does well is structural: examining your calendar for low-leverage hours, spotting an if-then incentive that will predictably narrow behavior, noticing that a change requires three handoffs, drafting 1:1 agendas that surface disengagement before it becomes a resignation. The judgment about a specific person stays with you — the diagnosis of the system around them is where this earns its place.
I’m a founder with no managers yet. Which of these is worth installing?
Drive Motivation and High Output Management. The first shapes how you design work and rewards before any bad incentive is embedded, which is far cheaper than removing one later. The second is really about leverage, and applies whether you’re managing five people or your own attention. Traction EOS and Team Topologies both assume structure you don’t have; installing them early tends to produce ceremony rather than clarity.
Every skill installs alone. The library installs together.
Install all four, or just one.
npx skills add wondelai/skills --all --global MIT licensed · Claude, Claude Code, Codex, Cursor, OpenClaw, Hermes · No account