Aimable Lab · Skills

A recipe your team and your agents can share

A Skill packages a prompt, the tools it uses, and the policies it inherits, into a named unit. Humans invoke it the same way agents do.

Live preview

Anatomy of a Skill

Read a contract, flag clauses against the house playbook, return a review memo.

review-contract
Shared Skill
v1.4
Owner
Legal team
What it does

Read a contract, flag clauses against the house playbook, return a review memo.

Inputs
  • file
    contract PDF or DOCX
  • collection
    playbook House playbook
Tools
clause.extractplaybook.searchpolicy.check
Guardrails
  • No external retrieval
  • EU data residency
  • Redact PII in outputs
Output
{ memo, clauses[], risks[] }
Invocation
Running
Caller
Aimee · agent
waiting for invocation…
One Skill, two audiences

Humans and agents run the same recipe

A Skill is defined once. A person invokes it by name in the chat. An agent calls it when the loop needs it. The Space, the tools, and the guardrails are identical either way.

  • Owners can version a Skill without breaking the callers.
  • Every invocation lands in the same audit trail.
  • Guardrails apply whether the caller is human or agent.
Yours, not theirs

Why a Skill beats a prompt you paste around

Hardcoded prompts and custom GPTs are a start. A Skill packages the tools and the policies with them, so it actually works in a regulated setting.

Hardcoded prompt
Aimable Skill
Tools
Whatever the caller happens to connect
Declared in the manifest
Policies
Up to the caller
Inherited from the Space
Versioning
Copy-paste drift
Owner publishes a new version
Invocation
Person at a keyboard
Person or agent, same call
Audit
Scattered across chats
One trail, one Space
Use cases

Skills people already run

Small, named, reusable. Each one a building block for the loop.

review-contract

Flag clauses against your playbook. Return a review memo.

score-deal

Score an incoming deal against your investment criteria. Return a memo and a score.

triage-ticket

Read a ticket, classify it, draft a reply, attach the policy that applies.

summarise-filing

Read a regulatory filing, flag new requirements, assign owners.

prep-meeting

Read the invite, pull the relevant Collection, draft a briefing.

close-out

Write a close-out memo from the Space's history at the end of a project.

How it works

Define, package, share

Three steps. No special IDE needed.

01

Define the recipe

Write the prompt, pick the tools, inherit the Space's policies. Declare the inputs and the output shape.

02

Package as a Skill

Name it, describe it, version it. The manifest lives in the Space.

03

Share with the Space

Every member and every agent in the Space can invoke it by name. Every call is in the audit trail.

Lab project

Early access for design partners

We work with partners on the first set of Skills, then open authoring more broadly. Want to help shape it?

Turn your best prompts into Skills.

Book a demo or tell us which workflow your team keeps repeating. We'll package the first Skill with you.