SkillHubuseskillhub.com
EN中文
Sign inOpen workspace
Developer PreviewPublic registry discovery and skill inspection are live. Runtime invocation requires an authenticated project key. Paid marketplace, payment capture, automated payouts, tax/KYC automation, and final email delivery remain prelaunch integrations.
Launch Preview

Skill discovery and manifest inspection are available today. Runtime execution is available for signed-in workspaces with Project Keys. Paid marketplace flows are still in preview.

View status
Agent infrastructure / skill runtime

Run agent skills like production infrastructure.

SkillHub is a registry and governance layer for reusable AI agent skills. Browse skills, inspect permissions and manifests, then run approved skills through controlled APIs or MCP.

SkillHub is in Launch Preview: discovery and inspection are live today; runtime execution is available behind signed-in Project Keys.

Browse SkillsView DocsPublish a Skill
Contract-first registryManifest, schema, and version checks
Project KeysPolicy gates before invocation
Operational proofStatus, latency, logs, and audit trails
Public registryAvailable

Search public Skills and compare safe marketplace metadata.

Manifest inspectionAvailable

Inspect schema, permissions, runtime target, and version state.

Runtime invocationRequires Project Key

Run REST / MCP calls only after workspace setup and policy checks.

Paid marketplacePreview

Paid listings, billing, refunds, and payouts stay visibly gated.

Launch Preview

What is a Skill?

A Skill is a reusable capability an AI agent can call, such as web research, data analysis, code execution, or document Q&A.

For builders

Browse reusable Skills and inspect manifests before adding them to your agent.

For publishers

Submit Skills with clear manifests, permissions, and review status.

For teams

Govern Project Keys, logs, limits, and runtime usage.

Review permissions, policies, and publisher checks before every agent call

Scopes visible

Permission review

See declared capabilities and risk level before a skill enters your project.

web.searchLowweb.fetchOKdata.storePassed
Policy checked

Runtime governance

Project keys, policies, logs, and limits keep runtime usage accountable.

Project Key requiredLowRate limit: 60 / minOKAllowed: web.search, web.fetchPassed
Review queue

Publisher review

Submission review separates ready skills from preview or blocked listings.

Static analysisLowPermission reviewOKSecurity scanPassed
View all skills

Browser Research

Verified

Research a web topic and return concise findings with source URLs.

researchbrowser
Runtime
HTTP
Permissions
3 scopes
Inspect

Dataset Summarizer

Submitted

Convert tabular data into structured notes, anomalies, and next actions.

dataanalysis
Runtime
HTTP
Permissions
3 scopes
Inspect

Code Runner

Verified

Execute code securely in isolated environments and return results.

devcode
Runtime
MCP
Permissions
4 scopes
Inspect

Doc Q&A

Submitted

Answer questions based on documentation and knowledge sources.

docsqa
Runtime
MCP
Permissions
1 scope
Inspect

How it works

1

Discover

Find the right Skill for your Agent.

2

Inspect

Review manifests, permissions, and runtime details.

3

Prepare Project Key

Create a workspace project and generate a Project Key.

4

Invoke

Call the skill via REST API or MCP with your key.

5

Monitor

Track usage, logs, and performance.

Start from a real skill contract

Explore the public registry first, then connect runtime only after project setup is ready.

Browse SkillsRead Docs