The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Pulsefeed X402 listing page.
Verify before you pay or install.
Two questions an agent has to answer before it acts, and neither is answered by a scanner that only looks at the present:
The second question is the one static scanners cannot answer. A rug pull is clean at review
time by construction: the package collects installs for weeks, and only then ships the patch
that runs code on npm i. Answering it requires yesterday's snapshot to exist, which is why
the series here starts on 2026-07-30 and cannot be reconstructed after the fact.
PulseFeed re-audits the whole MCP registry every night and diffs it against the previous day's snapshot, and re-probes the x402 endpoint population daily. Everything below is free, needs no key, and no account.
Hosted, no install:
Streamable HTTP. Also on the official MCP registry, Smithery and Glama.
Or run it locally — see mcp/:
Tools include check_x402_endpoint (is this endpoint safe to pay), mcp_check_server
(audit before installing), mcp_drift_check (the rug-pull check — pass your own
dependency list), mcp_security_report, x402_incidents and x402_changes.
Put it in your README. It states what changed in your package after people adopted it:
Use your registry name (io.github.you/your-server) or your npm package name.
A green badge is a public claim about your package, so it is only issued when the package is
actually in our snapshot. When it is not, the badge reads unwatched in grey — never
green. Reporting absence of measurement as evidence of cleanliness is a mistake we made once
publicly and will not repeat; see the correction.
Fails the build when something you already depend on changes dangerously:
Marketplace · source. Run it on a schedule, not only on pull requests — drift happens between your commits.
No key, CORS enabled, safe to call from a browser or a catalog page:
Subscribe without signing up — the filter lives in the URL, so there is no subscriber database and nothing to leak:
Full spec: /openapi.json.
We once published that 76% of x402 endpoints were dead. That measured our own parser, not the market, and the figure was corrected twice more after that — each time downward, each time for the same class of reason: our own behaviour recorded as somebody else's track record. The whole mechanism, every correction and the checklist that came out of it are kept at pulsefeed.dev/correction rather than quietly deleted.
If you find a number here that does not hold, open an issue — that page is where it will end up.
mcp/ | the MCP server (pulsefeed-x402-mcp on npm) |
lint/ | x402-payable — is your x402 endpoint actually payable |
data/ | published series and snapshots |
server.json | manifest for the official MCP registry |
MIT