EyeClaw

EyeClaw is the control cockpit for your OmniForces. Where the Workforce roster tells you who your AI workforce is, EyeClaw is where you open one OmniForce and manage how it operates: its personality, its memory files, the skills it may hold, its scheduled heartbeat, and the channels it is wired to.

Open EyeClaw from the desktop and select an OmniForce. The window is organized into five tabs:

• Soul — the OmniForce's personality and role configuration: who it is, how it speaks, what it is responsible for.
• Files — its business memory: the documents and notes this OmniForce carries with it.
• Tools — the Skill catalog: everything this OmniForce is able to hold, with each skill's risk level, credit price, and availability. This is also where you grant a skill to the selected Force, or take it back.
• Heartbeat — a schedule for this Force: choose a cadence and, optionally, one of its granted skills to line up for each beat. Every beat records its own outcome.
• Wirings — the channels and systems connected to this OmniForce (WhatsApp, email, calendar, CRM, and more).

Your current Neural Credits balance sits at the foot of the OmniForce list.

Skills are granted today; execution is not available yet

Nothing you do in the Tools or Heartbeat tabs runs a skill, and nothing there is charged. Granting records the capability against the selected Force and enforces the audit, risk, plan and credit gate at that moment. The executor — the part that takes a granted skill and performs the work — has not shipped. Until it does, no skill runs, no beat runs a skill, and your Neural Credit balance is untouched by both.

The prices below are the real per-run prices that will apply once execution ships. Skill Hub covers this in full.

Skills: audited, risk-rated, priced

As the app itself puts it: "Every Skill is Eyeleven-audited, risk-rated, and carries a Neural Credit price per run." Concretely, each skill carries four properties:

1. Audit status — only skills that have passed Eyeleven's audit can be granted. Skills still in the pipeline show In review.
2. Risk level — LOW, MEDIUM, HIGH, or CRITICAL, shown as a colored badge.
3. Credit price — per execution, shown on the skill (for example, Answer from Knowledge · 1 credit, Qualify Lead · 3 credits, Book Appointment · 5 credits, Generate Quote · 5 credits). It is what a run will cost once execution ships, not a charge you are incurring now.
4. Minimum plan — some skills require Nexus, Hyperion, or Omnis; on a lower plan the skill shows the plan it needs instead of a grant option.

Skills are granted to one named OmniForce at a time, not to your whole workspace — the Tools tab always acts on the Force you have selected. The full catalogue, with every skill's price and minimum plan, and what each refusal means, is documented in Skill Hub.

What a Force may do on its own

A skill is checked in a fixed order — audit, then risk, then your plan, then your credit balance — and the first check that fails is the one you are shown.

• Audit comes first. A skill that has not passed Eyeleven's audit shows In review, and no workspace can grant or run it whatever plan it is on. Send Payment Link and Charge Customer are both in review today, so that is the reason you see on them.
• Then risk. Skills rated HIGH or CRITICAL are not released to customer workspaces at all. There is no approval path for them, and no plan and no credit balance unlocks them. Bulk CRM Update is the one audited skill currently held back this way.
• Then your plan, and finally your credit balance. These last two are the only refusals you can resolve yourself, by upgrading or topping up.

A skill is therefore never "waiting for approval". Either the Force may run it, or it is refused with one of those four reasons. Skill Hub explains each one.

Approvals apply to control-plane commands

There is a human approval queue in Eyeleven, but it belongs to the commands issued from the EyeClaw control plane at eyeclaw.app — not to skills:

• Low-risk commands (reading summaries, listing records, checking status) run automatically.
• Medium-risk commands run automatically only when they are draft-only — they create drafts, previews, or analyses with no live side effects.
• Medium-risk commands with live side effects wait for approval. They are parked with the exact input they would run with until a workspace owner or admin releases or rejects them. Nothing executes meanwhile, and an approval can never unlock something the policy blocks outright.
• High- and critical-risk commands are blocked, not queued.

This is deliberate: an AI workforce should never take an irreversible action that no human authorized. See EyeClaw Control Plane.

How charging works

Today, skills are not charged at all — nothing runs, so nothing bills. Once skill execution ships:

• A skill's cost is charged when a run completes — after it has passed the audit, risk, plan, and credit checks and the work has been done.
• A skill refused by any of those checks is never charged. A blocked run costs you nothing.
• The cost applies per run, every run. Granting a skill costs nothing by itself, and that will not change.
• If your balance is below a skill's cost, the skill will not run until you top up or your cycle renews.

Heartbeat beats already report their own outcome honestly. A beat today reads "Beat recorded. The Skill was not run — skill execution is not available yet — and nothing was charged." When a beat cannot use its skill for another reason, it says which one — for example not enough Neural Credits, or the skill no longer being granted to that Force.

The control plane at eyeclaw.app

This cockpit configures one OmniForce. There is a second, separate surface at eyeclaw.app — the EyeClaw control plane — which is for operating the workspace as a whole: issuing commands, running several at once, orchestrating multi-step goals, deciding approvals, and exporting the full audit log.

It signs in with the same Eyeleven account. You do not need it to run your business — the desktop OS remains where you deploy Forces, build workflows, connect channels, and manage billing — but it is where an operator goes to drive and audit the workspace directly.

See EyeClaw Control Plane for what each of its surfaces does.

Getting started

1. Open EyeClaw, select your OmniForce, and read its Soul tab — adjust the personality before anything else.
2. In Tools, review which skills are available on your plan and what each will cost per run, then grant this Force the ones its role needs.
3. Wire the channels you need in Wirings.
4. Optionally set a Heartbeat cadence so the schedule is in place for when execution ships, and check its last-beat report the next day.