# agents.json

> A manifest that describes a site's APIs and workflows in a form agents can discover and execute, typically served from /.well-known/.

_The Agent Protocol Atlas · /protocols/agents-json · [JSON](/api/protocols/agents-json) · [all The Agent Protocol Atlas](/protocols)_

- **name:** agents.json
- **full_name:** agents.json / Arazzo API manifests
- **layer:** discovery
- **creator:** open community (built on the OpenAPI Arazzo spec)
- **status:** emerging
- **year:** 2025
- **one_liner:** A manifest that describes a site's APIs and workflows in a form agents can discover and execute, typically served from /.well-known/.
- **spec_url:** https://github.com/wild-card-ai/agents-json
- **snippet:**

```
GET /.well-known/agents.json → { "endpoints": [ ... ], "workflows": [ ... ] }
```
- **abbreviation:** agents.json
- **also_known_as:** agents.json, Arazzo API manifest
- **canonical_spec_url:** https://github.com/wild-card-ai/agents-json
- **entity_uri:** https://github.com/wild-card-ai/agents-json
- **taxonomy_layer:** discovery
- **sub_layer:** api-workflow-manifest
- **protocol_type:** manifest
- **central_problem:** Describes a site's APIs and multi-step workflows in a machine-readable manifest agents can discover and execute, rather than reverse-engineering the UI.
- **maintainer:** Wild Card AI + open community (built on the OpenAPI Arazzo spec)
- **governance_body:** none (community; layered on OpenAPI Arazzo)
- **license:** MIT
- **maturity_tag:** emerging
- **current_spec_version:** — (verify-against-primary-at-build)
- **spec_date:** — (verify-against-primary-at-build)
- **launch_date:** 2025
- **last_verified:** 2026-06-15
- **transport:** Static JSON manifest at /.well-known/agents.json (OpenAPI Arazzo workflows)
- **core_mechanism:** A JSON manifest served from /.well-known/agents.json enumerates a site's API endpoints and chained workflows using the OpenAPI Arazzo format, so an agent can discover which calls exist and execute multi-step flows without scraping the UI.
- **discovery_endpoint:** /.well-known/agents.json
- **settlement_type:** —
- **adoption_metric:** — (verify-against-primary-at-build)
- **notable_adopters:** {"value":"Wild Card AI","source":"https://github.com/wild-card-ai/agents-json"}
- **relationships:** {"predicate":"complements","target":"llms-txt","note":"agents.json declares APIs/workflows; llms.txt declares content — paired Layer-1 discovery files."}, {"predicate":"built_on","target":"mcp","note":"Both expose machine-callable actions; agents.json built on OpenAPI Arazzo and can front MCP-style execution."}
- **ideal_use_case:** Publishing a discoverable, executable description of your site's APIs and workflows for agents.
- **when_to_use:** When agents need to call your APIs or run multi-step workflows and you already have an OpenAPI surface to expose.
- **when_not_to_use:** When you only have content to expose (use llms.txt) or want a live tool-calling session rather than a static manifest (use MCP).
- **code_example:** GET /.well-known/agents.json HTTP/1.1

{ "endpoints": [ { "path": "/api/search", "method": "GET" } ], "workflows": [ { "id": "checkout", "steps": [ ... ] } ] }
- **source:** Spec/repo (built on OpenAPI Arazzo): https://github.com/wild-card-ai/agents-json ; agents.json/agents.txt listed among Layer-1 discovery additions in research §2.
- **agent_readiness_link:** agent-readiness/discoverability
- **layer_legacy:** content