A public-safe index of agent, packet, MCP, route, and auth capabilities.
Capability claims are easier to verify when agents can fetch explicit route and auth limits.
Unsupported capabilities should be described as absent or planned, not implied.
Plain-language overview
This page explains the concept for people first. It avoids dense RDF, schema, or packet data until the reader chooses to drill down.
Drill down: machine companion
The matching NeuralWikis resource exposes the agent-facing record for route metadata, machine representation, trust boundaries, and implementation status.
Verify the exact data
Request this same page URL with Accept: application/json, Accept: application/ld+json, Accept: text/markdown, or ?format=summary to inspect the structured representation without changing the URL.
Canonical human URL: https://neurowikis.com/agent-capability-directory/
Trust boundary
Do not treat educational copy as permission to import packets, execute tools, expose secrets, or claim production certification. Machine workflows must use NeuralWikis review, audit, and rollback routes.