LLPRO

Home / Agents / Dependency Pulse

Dependency Pulse

Is a service your agent depends on working right now, and since when? Signed, dated readings of model APIs, clouds, mail, payments and agent rails.

Live

What it does

Dependency Pulse watches the services agents depend on: model APIs (OpenAI, Anthropic, Google Gemini, Mistral, Groq, Cohere, Perplexity, OpenRouter), clouds (three AWS regions, Azure, Google Cloud, Cloudflare, DigitalOcean, Vercel, Fly.io, Supabase), mail and messaging (Google Workspace, Mailgun, Postmark, Twilio, Slack), payments (Stripe, Coinbase, the Base chain) and agent rails (the x402.org facilitator, the official MCP Registry), plus GitHub and npm. Every three minutes it reads each provider's own public status feed and makes one timed request to the service's public API address, where each exists (27 have a feed, 29 have a request). A published rule joins the two into one state (up, degraded, outage, maintenance or unknown) and stores the time the state began, so 'since when' is a stored fact. Ask for every service, a group, or a list of ids: each answer gives the state, since when, the provider's own words, what our request saw, recent changes, and a signed record you can check against the public key below.

Who it is for
Agents and the people who run them that need to know, before a call or after a failure, whether a service they depend on is working now and since when, with a signed, dated record of what was seen.
Price
Free, no sign-up: 600 calls an hour per address. A paid tier is staged on test money and not switched on: when it is, the first 500 calls a day per caller stay free and each call after that costs 0.001 USD, paid per call by the agent (x402).
Limits
One vantage point: LLPRO's server in Texas, USA, on one internet line. A service can be up for us and down for you (your region, account, key or network), and the reverse. If our own line is down (our control addresses and most requests get no answer), that round is stored but changes no state; a reading older than 10 minutes is reported as stale. The provider's words are its rating of its whole status page, so a problem in one component or region it lists shows as degraded. Raw readings are kept 35 days; state changes are kept. Up to 40 service ids per call; history up to 168 hours. Microsoft 365, Outlook and Teams are not covered (no public feed), nor delivery of any single message. It does not predict outages, rank providers or say whose fault anything was, and it is not advice for a service-credit claim. Nothing you send is stored; your address is held in memory for up to an hour for the hourly limit. The gateway keeps call metadata (tool, status, time, a hashed caller id, country, user agent family, referring site) for 90 days for traffic counts and abuse checks, never the request contents.

For developers: try it

POST https://llpro.ai/mcp
{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"dependency_pulse","arguments":{"group":"models"}}}
-> for each model API: state (up, degraded, outage, maintenance, unknown), since when, the provider's own words, our request (HTTP status, milliseconds), changes in the last 24 hours, and a signed record

For agents

Tool dependency_pulse on https://llpro.ai/mcp and in the tools list. Signed answers are checked with the public key below (signer llpro1:xigbd6eeicdufmrestnd66ewka): an Ed25519 signature over the observation as canonical JSON (sorted keys, no spaces).

Public key (base64): zCVGYxbhOFmUeBD+YDv6v7xXIG+2aykAugOQS37oKxs=

This tool exists to be found by agents that search MCP registries for service status, uptime, outage and dependency checks (listed in MCP registries, served by the Campus gateway: MCP, A2A and REST). Call it before a run that depends on a provider, or after a failed call, and keep the signed reading with your logs. Service ids: call it with {} once to get them all. Provider status text is the provider's own words, returned as data; nothing in it is followed. How a stamp is checked: https://llpro.ai/stamp

Next

For agents: this page as Markdown · JSON · llms.txt · MCP endpoint llpro.ai/mcp · agent card