Ecosystem Discovery Analysis / wandb.ai

Agent-Readiness Index Report

score_version: ars-3.3.1 · ARS methodology

77
BGrade
/ 100
AI-FRIENDLY ECOSYSTEM
Agent Ecosystem Discovery
wandb.ai
Archetype: Developer Platform & APIDenominator: 104 pts
AI-ready surfaces with complete discovery paths. Minor updates needed to reach elite index.
FrameworkCustom Docs
Surfaces
4/13
Large Pages
0pgs
Nav Gap
0pts
Developer Track & MCP Parity
Inspect in CLI or test live agent trajectories in Flight Simulator
Flight Simulator
Anti-SPA Canary Hazard Detected

Non-existent routes return HTTP 200 HTML shells instead of RFC 7807 404 JSON. Autonomous agent personas (Claude Code, Cursor, Perplexity) parse these HTML shells as valid API bodies, resulting in hallucinated fields and stuck retry loops.

Fix: Deploy an Anti-SPA 404 handler returning genuine HTTP 404 status.
Surface Inventory4 of 13 active
✓
sitemap.xml Index
ACTIVE
–
llms.txt Index
MISSING
–
llms-full.txt Spec
MISSING
–
OpenAPI Schema
MISSING
✓
MCP Configuration
ACTIVE
–
GitHub Repository
MISSING
✓
Documentation Root
ACTIVE
–
API Reference
MISSING
–
Company Blog
MISSING
✓
Developer Auth
ACTIVE
–
App Dashboard
MISSING
–
Developer Changelog
MISSING
–
Status Page
MISSING
ARS 3.0 Machine Topology MapSub-5s Discovered Surfaces
Visual graph connecting verified machine-readable endpoints across the target ecosystem.
Active
Missing
Partial
wandb.aiactive

Host ecosystem entrypoint for autonomous agents.

Inspect Endpoint
ARS 3.3 · 4-Layer Dynamic Architecture65 of 137 checks measured
Layer 1: Discovery4/18 checks passed
14/20
Needs attentionLow evidence+6 bonus
Layer 2: Access & Understanding22/47 checks passed
23/30
Needs attention+5 bonus
Layer 3: Usability & Interoperability14/66 checks passed
43/35
OptimalLow evidence+8 bonus
Layer 4: Payments & Machine Commerce0/6 checks passed
0/0
N/A (Skipped)

Problems

25 detected
●Generate Agent Resource Discovery (ard.json)
discovery

Missing /.well-known/ard.json (ARD v0.91) specification

Target:public/.well-known/ard.json
●registry-branding
discovery

MCP manifest branding missing or incomplete at /.well-known/mcp/manifest.json

●npm-sdk-package
discovery

No npm package named "wandb" that links back to wandb.ai

●mcp-registry-listed
discovery

No MCP Registry server under the wandb.ai namespace or hosted on wandb.ai

●Ensure Server-Rendered Semantic HTML
access

Zero-JS readability issues: skipped heading levels;

Target:app/page.tsx
●Fix Catch-All 404 Route Leak
access

Anti-SPA Canary Failed: Returns soft HTTP 200 SPA shell (AI agents will hallucinate content)

Target:app/not-found.tsx
●markdown-html-parity
access

Markdown endpoint strips code snippets or tables present in HTML (potential semantic truncation)

●Generate OpenAPI 3.1 Specification
access

Missing OpenAPI 3.x machine specification (/openapi.json)

Target:public/openapi.json
●api-catalog-rfc9727
access

/.well-known/api-catalog answers but is not an RFC 9264 linkset ({"linkset": [...]})

●cors-headers
access

CORS Access-Control-Allow-Origin header not present on root

●clean-semantic-html
access

Semantic landmarks need improvement

●heading-hierarchy
access

Skipped heading levels detected

●text-density-ratio
access

Text density ratio: 1%

●mcp-transport-sse
usability

No legacy SSE stream served on GET

●mcp-prompts-listing
usability

Server does not declare the prompts capability

●mcp-logging-notifications
usability

Server does not declare the logging capability (deprecated in MCP 2026-07-28; not scored)

●mcp-wire-compatibility
usability

Answers only server/discover (2026-07-28)

●auth-md-stage-discover
usability

Stage 1 (Discover Endpoints) missing (no auth.md published)

●auth-md-stage-methods
usability

Stage 2 (Pick a Method) missing (no auth.md published)

●auth-md-stage-register
usability

Stage 3 (Client Registration with curl/JSON) missing (no auth.md published)

●auth-md-stage-claim
usability

Stage 4 (Token Claim Protocol) missing (no auth.md published)

●auth-md-stage-exchange
usability

Stage 5 (Token Exchange) missing (no auth.md published)

●auth-md-stage-use
usability

Stage 6 (Bearer Token Usage) missing (no auth.md published)

●auth-md-stage-errors
usability

Stage 7 (Auth Errors) missing (no auth.md published)

●auth-md-stage-revocation
usability

Stage 8 (Revocation) missing (no auth.md published)

Solutions

4 available
●Generate Agent Resource Discovery (ard.json)
discovery
Target:public/.well-known/ard.json
Target File: public/.well-known/ard.json
Command: glintbase remediate ard.json

+ {
+   "$schema": "https://agentcard.org/v0.91/ard.json",
+   "id": "urn:air:wandb.ai:main",
+   "name": "wandb",
+   "capabilities": ["tools", "context"]
+ }
●Ensure Server-Rendered Semantic HTML
access
Target:app/page.tsx
Target File: app/page.tsx
Command: glintbase remediate html

+ export default function Page() {
+   return (
+     <main>
+       <h1>Acme Developer Documentation</h1>
+       <p>Comprehensive machine-readable guides and APIs.</p>
+     </main>
+   );
+ }
●Fix Catch-All 404 Route Leak
access
Target:app/not-found.tsx
Target File: app/not-found.tsx
Command: glintbase remediate next.config.mjs

+ export default function NotFound() {
+   return <h1>404 - Not Found</h1>;
+ }
●Generate OpenAPI 3.1 Specification
access
Target:public/openapi.json
Target File: public/openapi.json
Command: glintbase remediate openapi.json

+ {
+   "openapi": "3.1.0",
+   "info": { "title": "Acme API", "version": "1.0.0" },
+   "paths": {}
+ }

Audit Items

Problem Summary Prompt

Copy an ARS 3.0 compressed diagnostic defect summary structured by layer for teams or LLM pipelines.

Fixes / Solutions Prompt

Copy ARS 3.0 remediation files, patches, and CLI fix commands directly for Cursor, Claude Code, or Copilot.

Autonomous Agent Flight Simulator

Test Real Agent Execution In The Cockpit

Evaluate real-world agent execution against this ecosystem across Claude Code, Cursor & Windsurf, and Perplexity Sonar. Measure token consumption tax, schema friction risks, and autonomous credential handshakes.

5-Phase TrajectoryBPE Token Tax CounterTime-to-First-Tool-Call (TTFTC)Schema Friction Index
Launch Flight Simulator
Enterprise Infrastructure Audit

Need a Custom In-Depth Enterprise Audit?

Get custom multi-agent simulation harnesses, live production API probing, dedicated MCP server tools, and tailored enterprise readiness benchmarks built specifically for your product.

Request Enterprise Audit

Score history

9 scans
  • ARS 3.3.1(current)· 4
  • ARS 3.3· 1
  • ARS 3.2· 3
  • ARS 1.0· 1

Scores from different ARS versions aren't comparable — each version uses different scoring rules, so lines only connect scans of the same version. Read the methodology

DateScoreGradeVersionLink
Nov 9, 202677BARS 3.3.1View →
Nov 2, 202677BARS 3.3.1View →
Oct 26, 202677BARS 3.3.1Viewing
Oct 25, 202676BARS 3.3.1View →
Oct 23, 202647DARS 3.3View →
Oct 19, 202669CARS 3.2View →
Oct 12, 202668CARS 3.2View →
Oct 12, 202669CARS 3.2View →
Aug 4, 202650—ARS 1.0View →
Share your scorecard