Runtime

How Project Keys keep Skill runtime controlled

Project Keys separate public inspection from real invocation. Teams should scope keys by project, rotate them when owners change, and review logs after runtime calls.

Find related SkillsRequest custom workflow

What this page helps with

It turns a broad AI automation idea into a searchable, comparable, reviewable Skill workflow.

  • Clarify the job to be done
  • Choose the right Skill category
  • Separate public inspection from authenticated runtime

What to inspect before adoption

Do not evaluate a Skill by name only. Inspect its manifest, input/output schema, permissions, publisher, review state, support path, and version history.

  • Start with free basics for low-risk work
  • Use project policy for customer data or internal systems
  • Submitted Skills remain inspection-only until verified

Recommended next step

Start in the marketplace, open the Skill detail page, compare permissions and examples, then sign in only when the Skill should become project state.

  • Browse the marketplace
  • Read docs and install path
  • Contact SkillHub for custom workflows

Keep reading

Concept

What an AI Agent Skill is

An AI Agent Skill is a reusable capability with a manifest, input/output schema, permission profile, runtime entrypoint, publisher metadata, and review state. Treat it like infrastructure, not a prompt snippet.

Architecture

MCP and SkillHub are complementary

MCP helps agents discover and call tools. SkillHub adds marketplace discovery, publisher trust, permission review, project policy, runtime evidence, and commercial readiness around those tools.

How Project Keys keep Skill runtime controlled | SkillHub