Measurement log
Not a ratingMCP Heartbeat
An automated measurement log for remote MCP servers. It records whether a handshake succeeded, how long it took, whether the tool list changed, and what a cited document says about the free tier.
These figures are automated measurements with timestamps. They are not a review, rating, grade, or recommendation.
Measured tools
Search matches tool names, descriptions, and server names from live tools/list responses. Free-tier filter matches only records with a verified citation. Unknown stays unknown.
| Tool | Server | Measured | Latency | Free tier |
|---|---|---|---|---|
accept_job Accept a delivered job: it closes, no dispute is possible after it, and its secret is deleted. Without a dispute it counts as accepted 7 days after delivery anyway. The answer is short (job_id, status, accepted_at, li… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
answer_job_question Answer our team's open question on a job (status needs_info); the clock starts again. A login they ask for goes in secret (write-only), never in answer. | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
ask_human Ask your human owner a question by email (needs an ask-my-human utility from buy). Give 2 to 6 short options for one-tap answers, or free_text: true for a written answer. Plain words about a decision only: no links, i… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
audit_result_claim_x402 Paid deterministic audit for AI result claims. Price is $0.01 USDC on Base. Use the free trust receipt first when trust or official distribution-validation evidence is needed. After price approval, an unsigned call re… | ai.ai-akari/agent-trust-receipt | 2026-11-10T11:37:24.199Z | 12337 ms | unknown |
blueprint_delete_stage Delete one stage of your blueprint and every later stage (they are built on it), for good. Deleting tasks also deletes the task statuses. Works after the plan ended too. | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
blueprint_method The Project Blueprint method (free, no passport needed): six planning stages (concept, features, pages, database, tech_spec, tasks), each with instructions, a JSON Schema and a worked example. You plan with your own m… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
blueprint_put_stage Save one stage of your blueprint, in order: concept, features, pages, database, tech_spec, tasks. We check it against its schema and your earlier stages (the database DDL really runs on PostgreSQL). Problems come back… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
blueprint_tasks Mission control for your blueprint: every task with its status, the progress, and the next task to work on. | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
blueprint_update_task Set a blueprint task to doing when you start it and done when its tests pass (todo undoes). Your human sees the progress on the plan page. | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
buy Buy an item from your balance. Utilities start immediately; kits return a single-use download link. Pass instance_id to renew a utility you own. Above what is left of today's spending limit, the order waits (status pe… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
buy_site_badge Spends the balance: $5 for 30 days; renew with buy and the badge's instance_id ($3 per 30 days). Needs the domain proved with prove_domain in the last 7 days. We check the site daily and publish the result on a public… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
cancel_job Cancel a website test still waiting for its sites' proof (status needs_verification): the whole price comes back to your owner's balance as store credit at once, the job closes (closed_reason cancelled_unproved), and … | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
check_agent_one_page Checks the structure of a one-page agent summary written from Agent Blueprint's template (https://agentblueprint.ai/agent-one-page-template): the summary and all seven numbered sections present and in order, a stage l… | ai.agentblueprint/agent-blueprint | 2026-11-10T11:37:00.010Z | 500 ms | unknown |
check_emails Check up to 1000 of your owner's own email addresses before sending ($1 for up to 100 addresses, $4 for up to 1,000, counted by different entries; a list where no address needs the mail server costs $1, the rest credi… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
check_sender Check how the domain your owner sends email from is set up ($1 per domain, from the balance): MX, SPF, DKIM, DMARC, MTA-STS, TLS-RPT, BIMI, reverse DNS and a blocklist. The answer usually holds the whole report: a sco… | ai.ciphyr.agentmart/agentmart | 2026-11-10T11:37:02.735Z | 707 ms | unknown |
Latest probe attempts
| When (UTC) | Server | Outcome | Tools | Latency |
|---|---|---|---|---|
| 2026-11-10T11:37:26.278Z | AISTLabs ai.aistlabs/aistlabs | auth_required | — | 120 ms |
| 2026-11-10T11:37:24.199Z | AIﾉアカリ☆ Agent Trust Receipt ai.ai-akari/agent-trust-receipt | ok | 2 | 12337 ms |
| 2026-11-10T11:37:10.181Z | 1325.AI ai.1325/mcp | auth_required | — | 89 ms |
| 2026-11-10T11:37:08.540Z | DSAIL ai.jaxon/dsail | auth_required | — | 115 ms |
| 2026-11-10T11:37:06.514Z | FLORA ai.flora/flora | auth_required | — | 21 ms |
| 2026-11-10T11:37:04.918Z | AgentRoam ai.agentroam/agentroam | skipped_robots | — | — |
| 2026-11-10T11:37:02.735Z | AgentMart ai.ciphyr.agentmart/agentmart | ok | 112 | 707 ms |
| 2026-11-10T11:37:00.010Z | Agent Blueprint ai.agentblueprint/agent-blueprint | ok | 4 | 500 ms |
| 2026-11-10T11:36:57.847Z | Adrails ai.adrails/adrails | skipped_robots | — | — |
| 2026-11-10T11:36:56.119Z | Adako: ad ops for AI assistants ai.adako/ads | skipped_robots | — | — |
| 2026-11-10T11:36:54.294Z | Achromat Machine Operational Memory ai.achromat/mcp-server | skipped_robots | — | — |
| 2026-11-10T11:36:52.594Z | A1 Lane ai.a1lane/a1lane | auth_required | — | 637 ms |
Cited free-tier records: 0. Unknown: 2500. Calls in the last 7 days: 16. Distinct client hashes: 9.
Methodology
- Read the latest server records from the Official MCP Registry. Metadata submitted there is dedicated to the public domain under CC0.
- Keep remote streamable HTTP and SSE endpoints that publish no credential. Stdio packages are catalogued and not probed.
- Probe up to 200 eligible endpoints per cycle, spread across hosts. The order is a queue, not a quality score, and it is not published as one. On the public site each cycle is split into short batches.
- Send initialize, the initialized notification, and tools/list. Never tools/call. Wait between servers. Send an identifiable user agent. Skip paths disallowed by robots.txt.
- Store the protocol version, latency of the handshake, tool count, and a hash of each tool name, description, and input schema.
- Compare that hash with the previous successful probe. Added, removed, and changed tools are drift. The first success is a baseline, not a change.
- Free-tier rows start as unknown. A row changes only when a quote from the publisher's own page, or from registry metadata, is fetched and found on that page.
Limits
- A successful probe means the server returned a tool list. It does not mean the tools are safe, correct, or worth calling.
- Success rate is successful handshakes divided by connection attempts. Skipped robots and private URLs are not attempts.
- Tool descriptions can disagree with tool code. This service does not read source and does not call tools, so it does not measure that gap.
- Auth-walled servers are recorded as authentication required. They are not marked unhealthy for requiring a key.
- Unknown free tier means no citation is on file. It is not a claim that the server is paid or free.
- The registry can list a URL that later points somewhere else. Probes refuse local, private, and link-local addresses, then follow at most three public redirects.
Sources
Catalog rows come from the Official MCP Registry (registry.modelcontextprotocol.io), whose terms dedicate submitted registry metadata to CC0. This project is not the Official MCP Registry. Smithery's registry API requires an API key and publishes no reuse license that allows a copy of the catalog, so it is not ingested. Glama, G2, and Product Hunt are not fetched. Details are in docs/DATA_SOURCES.md.
Motivation, not a measurement from this service: an August 2026 snapshot of 21,643 registry servers (arXiv:2609.14119) found that 40.6% of multi-version servers changed description, package, or endpoint meaning while the registry name stayed fixed, and that the registry does not version tool definitions. Live tool-list hashes are the gap that paper points at.
Use it
The same data is a Streamable HTTP MCP server at https://mcpheartbeat.com/mcp and a JSON API. Anonymous callers are limited to 100 metered calls per UTC day per hashed IP. No account.
Cursor mcp.json
{
"mcpServers": {
"mcp-heartbeat": {
"url": "https://mcpheartbeat.com/mcp"
}
}
}Claude (remote MCP)
{
"mcpServers": {
"mcp-heartbeat": {
"url": "https://mcpheartbeat.com/mcp"
}
}
}VS Code, ChatGPT, or any Streamable HTTP client
https://mcpheartbeat.com/mcpJSON API
Every response is labelled automated measurements and includes a timestamp and Official MCP Registry attribution. Anonymous callers share one limit of 100 metered calls per UTC day per hashed IP. GET /api/v1/stats adds catalog counts, a 7-day summary, and 30 days of calls and distinct clients for /mcp and the API. A distinct client is the hashed IP plus the MCP client name, or the X-Client-Name header on the API.