◎ Discry Score
arcade.dev.com
infrastructure · API
A
0 / 100
DISCOVERY0
COMPREHENSION0
Category leader: 98 (A)
Discry your API →
INFRASTRUCTURE · RANK #3 OF 68

Arcade.dev is Agent-Ready to agents.

Discry independently scored how well an AI agent can discover and understand the Arcade.dev API from what’s public — not whether it’s usable. Below: every signal we checked, what’s costing the score, and what to change.

Discry your API — freeView the docs ↗

SCORED UNDER RUBRIC 1.2 · A full re-launch under Discry Score 2.5 — a new behavioral instrument, not comparable to these scores — is in progress.

Discovery

45% of score · 95/100
OpenAPI specA machine-readable OpenAPI/Swagger spec agents can parse.Pass
llms.txtAn llms.txt index that points agents to the docs that matter.Pass
llms.txt qualityThe llms.txt is focused, current, and well under the size budget.Pass
llms-full.txtA full-text bundle agents can load in one request.Pass
AGENTS.mdAn AGENTS.md that tells coding agents how to build on the API.Pass
.well-known/mcp.jsonA discoverable MCP manifest at a well-known path.Fail
MCP registryThe API is listed in a public MCP registry.Pass
robots.txt AI directivesrobots.txt allows (or explicitly guides) AI crawlers.Pass
SitemapA sitemap so agents can enumerate the docs surface.Pass

Comprehension

55% of score · 96/100
Task-oriented descriptionsEndpoints described by what they accomplish, not just their shape.Pass
Realistic examplesRunnable, real-world request/response examples.Pass
Multi-step workflowsDocs that chain calls into complete jobs an agent can follow.Pass
Error-recovery guidanceDocumented failure modes and how to recover from them.Pass
Answer-first formatThe answer leads; preamble does not bury it.Pass
Capability boundariesClear limits — what the API can and cannot do.Partial
Naming consistencyConsistent, predictable naming across endpoints.Pass
Heading hierarchyClean heading structure agents can navigate.Pass
Markdown docsDocs available as clean markdown, not JS-rendered HTML only.Pass
Token efficiencyDocs are concise enough to fit an agent context window.Pass

What we found

  • An agent (or its crawler) that lands on arcade.dev is fully welcomed: robots.txt explicitly Allows GPTBot, ClaudeBot, CCBot, PerplexityBot, Google-Extended and more, and adds Content-Signal: ai-train=yes, ai-input=yes — a deliberate, agent-first discovery posture rather than the default-block most sites ship.
  • A coding agent looking for Arcade would find it everywhere it looks: root llms.txt (concise, product/API-focused), a full docs llms.txt index with per-page descriptions, a 9.3KB llms-full.txt, a public OpenAPI 3.0 interactive reference (reference.arcade.dev, base api.arcade.dev), AGENTS.md files in the docs and create-arcade-agent repos, and MCP-registry listings on Glama.
  • Comprehension is best-in-class: the quickstart teaches a real multi-step workflow (search Google News -> create a Google Doc -> email the link) with copy-pasteable, realistic examples in Python, TypeScript AND Java — an agent can chain tools correctly on the first read.
  • Error recovery is written FOR agents, not just humans: a dedicated error-handling section documents RetryableToolError and ContextRequiredToolError with additional_prompt_content fed back to the LLM, plus 429 -> UpstreamRateLimitError and typed network/transport failures. This is the self-correction guidance most APIs lack entirely.
  • The only real discovery gap is a missing /.well-known/mcp.json — surprising for a company that co-authored the MCP authorization spec — and AGENTS.md is absent from the flagship arcade-mcp/arcade-ai SDK repos (present only in docs and the scaffolding CLI).

What to change

Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.

  1. 01Add /.well-known/mcp.json declaring Arcade's hosted MCP tools and auth model — the one near-universal discovery signal Arcade is missing, and a natural fit given Arcade is an MCP runtime authored by MCP-spec contributors.
  2. 02Add AGENTS.md to the flagship SDK repos (ArcadeAI/arcade-mcp and ArcadeAI/arcade-ai), not just the docs and create-arcade-agent repos, so coding agents working directly against the SDKs get the same in-repo guidance.
  3. 03Expose the OpenAPI 3.0 spec at a stable, machine-readable path (e.g. reference.arcade.dev/openapi.json or api.arcade.dev/openapi.json) — the interactive reference is public but the raw spec file isn't discoverable at a conventional URL, so agents can't fetch the schema directly.
  4. 04Publish an explicit capability-boundaries / rate-limits page with concrete numeric quotas and header names for the Arcade Engine API itself (today limits are documented mainly as error behavior), so agents know limits before they hit them.
  5. 05Document pagination and idempotency-key support for Engine API list/execute operations so long-running agents can page results and safely retry without duplicate side effects.

Execution coverage · INFORMATIONAL, UNSCORED

Whether an agent can actually complete a call and recover from errors is the deeper Audit layer — documented here, but not part of the Discry Score.

API key (ARCADE_API_KEY)OAuth 2.0 (per-user downstream tool authorization)SSO/IdP (Auth0, Okta, Microsoft Entra ID, Clerk, Stytch)RBAC✓ Error format documented✓ Rate limits documented✕ Pagination documented✕ Idempotency documented

Execution documentation is exceptional for agent use. Arcade documents a structured, machine-parseable error hierarchy (ToolError -> ToolRuntimeError -> RetryableToolError / ContextRequiredToolError / UpstreamError / UpstreamRateLimitError / NetworkTransportError) with status_code and ErrorKind classification, plus automatic error adapters that translate httpx/requests/SDK exceptions into typed Arcade errors. Auth is first-class: API-key for the Engine plus per-user OAuth2 authorization flows and pluggable IdPs. Rate-limit handling is documented via UpstreamRateLimitError (429), though explicit numeric platform quotas are not prominently published. Pagination and idempotency keys are not clearly documented. Base URL is https://api.arcade.dev with a public OpenAPI 3.0 interactive reference at reference.arcade.dev.

See your own Discry Score.

Drop your API docs URL. See what an agent sees — in 60 seconds, free.

Discry your API — free