A2A Agent Cards as First-Class Capabilities

Google's Agent-to-Agent (A2A) protocol describes agents and their skills with an Agent Card. ASPL reads that card and turns each skill into a capability your agents can discover and verify.

From card to capabilities

POST /v1/ingest/a2a
{ "name": "TranslatorAgent",
  "url": "https://a2a.example",
  "skills": [
    { "id": "tr", "name": "translate",
      "description": "translate text between languages" } ] }

Each skill is converted into an ASPL capability tagged with its A2A origin. The agent's identity is captured as a passport, and the skills become individually discoverable by intent.

Why this is useful

A2A tells you an agent claims a skill. ASPL adds the layer A2A doesn't: a trust score earned from real, distinct usage; a signed, hash-verified handoff when the capability is acquired; and a revocation path if the skill turns out to be harmful. You keep using A2A for the actual agent-to-agent interaction — ASPL just makes the discovery and the vouching trustworthy.

One fabric, many sources

Because MCP tools, A2A skills, LangChain tools, and natively-published capabilities all become the same kind of object, a single need query searches across all of them. Your agent doesn't have to know or care that one match came from an MCP server and another from an A2A card — they're ranked together by intent and trust.

ASPL is deliberately a layer, not a silo: it federates the ecosystems you already use rather than asking you to choose one.