Design partner — not shipped

Governance for AI agent skills, before it exists

Asterism does not enforce skill policies, gate installs on approval, or write an audit record today. We want to build that layer with one team that actually has the problem, rather than guess at it.

Below is what we intend to build. Treat all of it as unbuilt.

Skill Policies (planned)

Whitelist or blacklist skills per organization, with a minimum security score, enforced at install time.

Approval Workflows (planned)

Require review and approval before a skill can be installed.

Audit Trails (planned)

A durable log of every skill installed, updated, or removed — by whom, when, and into which harness.

Private Catalogs (partly built)

Private skills and org namespaces exist. Curating them into an enforced internal catalog does not.

What a design partner gets

- Direct access to the people writing the code
- The enforcement layer built against your actual policy
- An honest answer about timelines, including “not soon”

Request early access

Tell us about your team and we'll reach out to get you set up.