Resource types
Skills, rules, hooks, tools, and open-source projects are documented by what they do and how they affect a workflow.
Agent Exchange
Explain resource types, approval states, safety notes, and usage boundaries for humans and agents.
Skills, rules, hooks, tools, and open-source projects are documented by what they do and how they affect a workflow.
Every listing should keep upstream source, license, install behavior, outputs, and safety notes visible.
Human browsing and agent-facing discovery stay separate through approval state and user policy.
Teams still test resources in low-risk workspaces before using them with important code or systems.
Approval boundary
Trust labels explain review posture; policy still determines what a specific agent workflow can see.
A listing can be browsed by people for evaluation without being exposed to agents.
A listing can appear in agent-facing discovery only after approval and unresolved-report checks.
Open concerns keep access conservative until review work resolves the issue.
A user or team policy can narrow approved resources to an allowlist or blocklist.
Manifest flow
Manifests organize evidence for review; they do not remove the need to inspect source or test behavior.
01
Start with public listing context and compare similar resources.
02
Read source, license, install, outputs, and usage guidance in a structured format.
03
Run a small non-sensitive workflow before relying on the resource.
04
Expose only the resources that fit team policy and approval state.
Docs routes
Browse reusable AI resources and filter them by type.
OpenCompare selected resources side by side.
OpenSee which resources people can view and which AI helpers can access.
OpenReview read-only discovery and manifest endpoints.
OpenGet help with a listing, source, license, or review status.
OpenReport a sensitive vulnerability or abuse concern.
OpenReturn to the Agent Exchange overview.
Open