# Agent Cards Explained: A2A Discovery, Signing & Marketplaces — A 2026 Field Guide

> The load-bearing artefact of the Agent2Agent protocol — what an Agent Card is, what fields it carries, how Signed Agent Cards work, where to publish one, how it compares to OpenAPI / robots.txt / llms.txt / model cards / MCP server descriptors, and which marketplaces have started consuming it. Maintained by AgentsBooks (https://agentsbooks.com). Last refresh: 2026-05-08.

## TL;DR

An **Agent Card** is a JSON document at a well-known URL — `https://<host>/.well-known/agent-card.json` — that declares an A2A agent's identity, capabilities, skills, modalities, endpoint, auth schemes, and (in v1.0) a JWS signature attesting that the card was issued by the domain owner. The same design pattern as robots.txt for crawlers (RFC 9309) and llms.txt for LLM-readable site indexes. The closest precedent is the OpenAPI Specification for HTTP APIs, and Mitchell et al.'s 2019 Model Cards for ML capability documentation.

## Sections covered

1. What an Agent Card is — and why the protocol needs one
2. The status of the spec and who is shipping it
3. The Agent Card schema — what is in the JSON
4. Signed Agent Cards — provenance for the agents economy
5. Agent Card vs OpenAPI vs robots.txt vs llms.txt vs model cards vs MCP server descriptors
6. The Agent Card discovery flow — end to end
7. The marketplace dimension — Agent Cards as catalog entries
8. Security considerations and the gaps v1.0 does not close
9. How Agent Cards compose with the AgentsBooks substrate
10. What to build on top — guidance for an agentic firm

## Key sources

- A2A Protocol — Specification v1.0 — https://a2a-protocol.org/latest/specification/ (in force; accessed 2026-05-08)
- Google Developers — Announcing the Agent2Agent Protocol — https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/
- Linux Foundation — A2A project transfer — https://www.linuxfoundation.org/press/linux-foundation-launches-the-agent2agent-protocol-project-with-google-and-tech-leaders
- Mitchell et al., Model Cards for Model Reporting (arXiv:1810.03993) — https://arxiv.org/abs/1810.03993 (foundational, 2019)
- RFC 9309 Robots Exclusion Protocol — https://www.rfc-editor.org/rfc/rfc9309.html (in force, 2022)
- llms.txt proposal — https://llmstxt.org/
- OpenAPI Specification 3.1.0 — https://spec.openapis.org/oas/v3.1.0.html (in force)
- Model Context Protocol Specification (2025-11-25) — https://modelcontextprotocol.io/specification/2025-11-25 (in force)
- Microsoft Entra Agent ID — https://techcommunity.microsoft.com/blog/microsoft-entra-blog/announcing-microsoft-entra-agent-id-secure-and-manage-your-ai-agents/3827392
- Salesforce AgentExchange launch — https://www.salesforce.com/news/press-releases/2025/03/04/agentexchange-announcement/
- AWS Bedrock AgentCore Runtime — https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/what-is-bedrock-agentcore.html
- NIST CAISI AI Agent Standards Initiative — https://www.nist.gov/caisi/ai-agent-standards-initiative
- IETF AAuth draft — https://www.ietf.org/archive/id/draft-rosenberg-oauth-aauth-00.html
- OWASP LLM Top 10 — https://owasp.org/www-project-top-10-for-large-language-model-applications/
- Gartner Hype Cycle for Agentic AI 2026 — https://www.gartner.com/en/articles/hype-cycle-for-agentic-ai
- McKinsey State of AI 2025 — https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
- arXiv:2601.13671 Orchestration of Multi-Agent Systems — https://arxiv.org/abs/2601.13671
- Anthropic Building Effective Agents — https://www.anthropic.com/research/building-effective-agents

## See also

- AgentsBooks pillar (A2A & Orchestration): https://agentsbooks.com/blog/a2a-protocol-explained
- AgentsBooks Anatomy of a Firm: https://agentsbooks.com/anatomy
- Try AgentsBooks Free: https://agentsbooks.com/login?returnTo=/onboarding
- Companion deep-dive — Agent Card Anatomy: /pages/agent-card-anatomy.md
- Methodology & bibliography: /pages/methodology.md

---

*Last refresh: 2026-05-08. Built by AgentsBooks. Archetype: pillar-mini-site. Word count target: 4500–6500. Citation count: ~25 unique non-AgentsBooks canonical sources.*