together.ai
Journey
81.2/ 100
B · Ready with gaps
Waymark scored together.ai 81.2/100 (B) on 24 Aug 2026.
together.ai is usable with gaps. Usability is the strongest layer, and Discovery is the open one. Next: cLI tool.
FindMixed
14.8/20
12 of 18 checks passed
DoorMixed
24.7/30
17 of 24 checks passed
ToolsMixed
33.5/40
9 of 12 checks passed
Pay
N/A
No cart or pay rail — out of the denominator.
Group by
1. Can an agent discover and trust you? 68/100
Whether an agent can find this origin from public files and treat it as the real one.
pass
HTTPS reachable
2/2
Evidence. Homepage responded 200 over HTTPS.
Probed homepage.
pass
robots.txt present
2/2
Evidence. Valid robots.txt with User-agent rules.
Probed robots.txt.
pass
XML sitemap
2/2
Evidence. Found sitemap.xml or sitemap index.
failed
Well-known AI catalog
0/1
Evidence. No ai-plugin.json or ai-catalog.json.
Recommended fix. Add /.well-known/ai-plugin.json or /.well-known/ai-catalog.json.
failed
AGENTS.md
0/1
Evidence. agents.md status 404.
Probed public URLs named in the evidence.
Recommended fix. Publish /agents.md with agent-oriented usage notes.
pass
robots.txt AI policy
Required
2/2
Evidence. Named AI crawlers: Google-Extended.
pass
Sitemap for agents
Recommended
2/2
Evidence. Sitemap document fetched.
Probed sitemap.
failed
HTTP Link header
bonus
0/1
Evidence. No rel=sitemap|describedby|api-catalog|alternate on homepage.
Probed homepage + sitemap.
Recommended fix. Send RFC 8288 Link headers for related machine resources.
failed
ai-plugin.json
bonus
0/1
Evidence. No /.well-known/ai-plugin.json.
Probed well-known paths.
Recommended fix. Optionally publish a plugin-style capability manifest.
failed
API catalog linkset
bonus
0/2
Evidence. API present without RFC 9727 catalog.
Recommended fix. Add /.well-known/api-catalog pointing at OpenAPI.
scanner error
Wikipedia sitelink
bonus
0/4
Evidence. Wikidata SPARQL The operation was aborted
Scanner error — not counted as a site fail and not in the score denominator.
scanner error
Wikidata official website
bonus
0/3
Evidence. Wikidata SPARQL The operation was aborted
Scanner error — not counted as a site fail and not in the score denominator.
scanner error
Brand official-website match
Recommended
0/3
Evidence. Wikidata SPARQL The operation was aborted
Scanner error — not counted as a site fail and not in the score denominator.
pass
Public MCP registry listing
bonus
1/1
Evidence. Registry hit at https://registry.smithery.ai/servers?q=together.
Probed public URLs named in the evidence.
pass
Package homepage match
bonus
1/1
Evidence. npm package homepage matches together.ai.
Probed homepage.
pass
Sitemap lastmod
bonus
1/1
Evidence. Sitemap has lastmod.
n/a
related-website-set
bonus
0/1
Evidence. No related-website-set (N/A unless multi-origin).
n/a
ads.txt
bonus
0/1
Evidence. No ads inventory (N/A).
2. Do you welcome agents? 38/100
Whether crawlers and signed bots are allowed through rather than blocked at the door.
failed
AI crawlers not blanket-blocked
0/2
Evidence. robots.txt Disallow: / for * or common AI bots.
Probed robots.txt.
Recommended fix. Avoid Disallow: / for GPTBot, ClaudeBot, PerplexityBot unless intentional.
n/a
robots Crawl-delay
bonus
0/1
Evidence. No Crawl-delay (optional).
n/a
Web Bot Auth keys
Recommended
0/2
Evidence. No http-message-signatures-directory.
partial
CORS / API hint
0.5/1
Evidence. No CORS header or obvious API path.
Recommended fix. If you expose an API, send Access-Control-Allow-Origin for intended clients.
pass
CSP or X-Frame-Options
1/1
Evidence. CSP or X-Frame-Options present.
3. Does an agent understand who you are and what you do? 94/100
Whether the site explains the product in language a model can ingest.
pass
Title and description
2/2
Evidence. Title and meta description present.
pass
JSON-LD structured data
4/4
Evidence. application/ld+json script found.
pass
html lang attribute
1/1
Evidence. html[lang] is set.
pass
JSON-LD entities
Recommended
4/4
Evidence. Recognized schema.org types in JSON-LD.
pass
JSON-LD sameAs
bonus
2/2
Evidence. sameAs present.
pass
Open Graph basics
bonus
2/2
Evidence. Open Graph plus canonical.
pass
llms.txt
2/2
Evidence. Found /llms.txt with content.
pass
llms.txt index
Recommended
2/2
Evidence. llms.txt has content.
pass
llms.txt size and links
Recommended
2/2
Evidence. llms.txt 39088 chars, 543 URL(s).
failed
llms-full.txt
bonus
0/1
Evidence. llms-full.txt status 404.
Probed public URLs named in the evidence.
Recommended fix. Optionally publish /llms-full.txt for one-shot ingest.
n/a
Section llms.txt
bonus
0/1
Evidence. No /docs/llms.txt.
n/a
NLWeb schemamap
bonus
0/1
Evidence. No schemamap (N/A unless you offer NL search).
pass
Trust pages
Recommended
2/2
Evidence. Trust links found.
n/a
Speakable markup
bonus
0/1
Evidence. No SpeakableSpecification (N/A unless voice content).
n/a
humans.txt
bonus
0/1
Evidence. No /humans.txt (optional).
pass
security.txt
Recommended
1/1
Evidence. RFC 9116 security.txt with Contact.
pass
RSS or Atom feed
1/1
Evidence. Feed link detected.
partial
RSS or Atom document
bonus
0.5/1
Evidence. No /feed or /rss.xml document.
Recommended fix. Publish /rss.xml or /atom.xml.
4. Can an agent integrate with you? 80/100
Whether there is a documented machine door into the product.
failed
OpenAPI / Swagger link
0/2
Evidence. No OpenAPI/Swagger link in HTML.
Probed OpenAPI.
Recommended fix. Link to an OpenAPI or Swagger document from the homepage.
pass
Login or docs link
3/3
Evidence. Login or developer docs link found.
pass
MCP well-known card
2/2
Evidence. MCP server card found under .well-known.
Probed well-known paths.
failed
MCP SSE / HTTP hint
0/1
Evidence. No MCP/SSE hint on the homepage.
Probed homepage.
Recommended fix. Document an MCP HTTP or SSE endpoint.
n/a
openapi.json fetchable
0/0
Evidence. Fetched OpenAPI or Swagger document.
Probed OpenAPI.
pass
Webhook or API docs
2/2
Evidence. Webhook or API docs mention found.
n/a
Agent skills index
0/2
Evidence. No agent-skills index (N/A unless you publish skills).
pass
OpenAPI document
Required
5/7
Evidence. OpenAPI probe.
Probed OpenAPI.
pass
Public API surface
Recommended
7/7
Evidence. Docs or API paths look callable.
pass
Developer portal
Recommended
6/6
Evidence. Developer hub found.
pass
MCP server
Recommended
6/6
Evidence. MCP probe.
pass
MCP server card
Recommended
2/2
Evidence. MCP card JSON found.
n/a
Agent Skills files
bonus
0/2
Evidence. No SKILL.md surface (N/A).
failed
CLI tool
Recommended
0/3
Evidence. No CLI mention.
Recommended fix. Publish an official CLI.
n/a
GraphQL GET hint
bonus
0/2
Evidence. No GraphQL surface (N/A).
n/a
OpenAPI UI page
bonus
0/1
Evidence. No swagger-ui/redoc page.
n/a
OpenAPI (legacy slot)
0/0
Evidence. OpenAPI/Swagger fetched.
Probed OpenAPI.
5. Is your integration well-built? 100/100
Whether the integration surface is complete enough to call with confidence.
pass
MCP initialize
Required
3/3
Evidence. initialize HTTP 200 at https://togetherai-52386018.main-kill-isr.mintlify.me/mcp
Probed public URLs named in the evidence.
pass
MCP server card
Recommended
2/2
Evidence. MCP card JSON found.
6. Can an agent use you reliably in production? 100/100
Whether errors, idempotency, and limits are visible before a call fails.
pass
Rate limit signal
Recommended
2/2
Evidence. Rate-limit or retry signal found.
pass
Idempotency signal
Recommended
3/3
Evidence. Idempotency mention found.
pass
Machine-readable errors
Recommended
4/4
Evidence. JSON/problem error shape found.
pass
Honest 404
Recommended
2/2
Evidence. 404/410 on probe path.
7. Can an agent authenticate to you? 63/100
Whether an agent can obtain access without a human in the loop.
failed
OAuth / OIDC well-known
0/2
Evidence. No oauth-authorization-server or openid-configuration.
Recommended fix. Publish /.well-known/openid-configuration or oauth-authorization-server if you authenticate agents.
n/a
OAuth 2.0 surface
Required
0/5
Evidence. No delegated-access surface (N/A).
n/a
OAuth AS metadata
Required
0/3
Evidence. No AS metadata (N/A unless you run an AS).
n/a
PKCE S256
Required
0/2
Evidence. No OAuth metadata to evaluate PKCE.
n/a
OAuth protected resource
Recommended
0/2
Evidence. No OAuth resource metadata (N/A).
pass
Scoped permissions
Recommended
5/5
Evidence. Named scopes found.
n/a
auth.md
bonus
0/2
Evidence. No /auth.md (optional).
failed
change-password well-known
bonus
0/1
Evidence. Login surface without /.well-known/change-password.
Probed well-known paths.
Recommended fix. If users have passwords, serve /.well-known/change-password.
8. Can an agent transact with you? 100/100
Whether an agent can price, pay, or check out on a documented rail.
n/a
x402 payment
bonus
0/0
Evidence. No cart/SKU/checkout and no live payment rail (Pay N/A).
n/a
Agentic checkout (ACP)
bonus
0/0
Evidence. No cart/SKU/checkout and no live payment rail (Pay N/A).
n/a
ACP delegated payment
bonus
0/0
Evidence. No cart/SKU/checkout and no live payment rail (Pay N/A).
n/a
Universal Commerce Protocol
bonus
0/0
Evidence. No cart/SKU/checkout and no live payment rail (Pay N/A).
n/a
Machine Payments Protocol
bonus
0/0
Evidence. No cart/SKU/checkout and no live payment rail (Pay N/A).
pass
Pricing page
Recommended
3/3
Evidence. Fetched /pricing.
Probed public URLs named in the evidence.
n/a
pricing.md
bonus
0/2
Evidence. No /pricing.md (optional).
9. Can a user act through an agent? 0/100
Whether a person can finish a job through an agent-facing UI.
failed
MCP Apps UI hint
bonus
0/1
Evidence. No MCP Apps hint (N/A unless you return UI).
Recommended fix. If you return UI, declare io.modelcontextprotocol/ui resources.
n/a
A2A agent card
Required
0/2
Evidence. No agent-card.json (N/A unless A2A).
n/a
Web app manifest
bonus
0/1
Evidence. No web app manifest (N/A).
10. Can an agent operate your website directly? 67/100
Whether an agent can read and operate the site itself when there is no API or MCP.
pass
Markdown negotiation
bonus
1/1
Evidence. Accept markdown returned text/markdown; charset=utf-8.
failed
/index.md fallback
bonus
0/1
Evidence. index.md status 404.
Probed public URLs named in the evidence.
Recommended fix. Serve a markdown homepage at /index.md.
n/a
WebMCP page tools
bonus
0/1
Evidence. No in-page WebMCP (N/A).
n/a
NLWeb /ask
bonus
0/1
Evidence. No NLWeb /ask (N/A).
pass
Agent mode view
bonus
1/1
Evidence. mode=agent responded.
Probed public URLs named in the evidence.
n/a
NodeInfo
bonus
0/1
Evidence. No nodeinfo (N/A).
n/a
oEmbed discovery
bonus
0/1
Evidence. No oEmbed (N/A unless embeddable).
About this snapshot & how to cite it
Public technical signals observed by Waymark. A score describes this snapshot; it does not guarantee that an agent will complete a task. Private and authenticated workflows are outside this scan.
Snapshot: . Cite the domain, score, scan time, and report URL. This URL serves the latest completed snapshot and can change after a rescan.
Share this snapshot
Also on the Board
- tavily.com85/100 · BScan them →
- pinecone.io84.9/100 · BScan them →
- composio.dev82.3/100 · BScan them →
Email me if this score changes
Join the monitoring waitlist. Monitoring is not active yet. This saves your interest; it does not schedule scans or send change alerts.
Email this report
Send the current snapshot and a link. One email. Not a list.