Layer 01
Canonical capability manifests
The outer boundary that makes the rest reviewable.
Agents, MCP & capabilities · Concept
PluginRegistry, specified end to end.
A trusted machine-readable registry for discovering, validating and managing interoperable AI plugins, skills, tools and agent capabilities.
01
Positioning
One sentence, then the boundaries that make it credible.
A trusted machine-readable registry for discovering, validating and managing interoperable AI plugins, skills, tools and agent capabilities.
Status: Concept. The product is documented as a specification; production behaviour is not claimed.
02
Scope
What the brief commits to, and what it explicitly does not.
- Who it is for
- Developers, enterprises, platform providers, software publishers
- What it replaces
- unverified plugin directories, fragmented registries and manually maintained integration catalogs
- What it produces
- A controlled surface that humans and agents share.
- What it refuses
- Unscoped access, silent automation and claims without evidence.
03
Shape
Three layers, ordered from boundary to contract.
Layer 02
Identity and provenance verification
The outer boundary that makes the rest reviewable.
Layer 03
Compatibility and conformance testing
The outer boundary that makes the rest reviewable.
04
Status of this site
The same disclosure appears on every page.
PluginRegistry is classified Concept in the inventory. Code or reference work exists for part of the system; it is not a deployed service. This site describes the intended product, its interfaces and the evidence a release would have to produce.
No customers, benchmarks, certifications or production deployments are claimed anywhere on this site.