# Datasource Pulse

> Datasource Pulse health-checks any HTTP endpoint — anything with a URL can be monitored, and the 35+ pre-wired vendor list (OpenAI, Anthropic, GitHub, Stripe, SerpAPI and more) is a head start, not a limit. It watches API credentials AND usage quotas across any endpoint, plus the output quality of third-party Apify Actors you don't own, then sends exactly one deduplicated alert when something degrades and one notice when it recovers. Every verdict carries a machine-readable failure class agents can branch on, and agents can buy a single check live via x402 for $0.02 with no account. Runs on a schedule you choose; billed per connector checked.

Last updated: 2026-08-06

## Why it's different

- No ceiling: any HTTP endpoint works via one generic_http check — your own API, a partner feed, your database's REST or health endpoint. The 35+ pre-wired vendors are a head start (each vendor's credential and quota quirks already handled), not a limit.
- Credentials AND quotas in one probe: proves the token is alive and reads current usage against your own soft caps, across any endpoint — not just key validation.
- Monitors Actors you don't own: schema drift, volume drops, staleness, and cost spikes on third-party scrapers whose maintainers you can't control.
- One alert per state change: a deduplicated alert when a connector starts failing, one recovery notice when it heals. No alert storms.
- Machine-readable failure classes: a stable failure_class enum (dead_token, schema_drift, rate_limited, quota_exhausted, ...) agents branch on without parsing human error strings.
- Live x402 payability: an agent can buy a health check today, on-chain, with no account — see below.

## What it does

- Probe type 1 (credentials & quotas): checks that API keys are still valid and quotas aren't about to run out (e.g. an OpenAI key, a GitHub personal access token, a SerpAPI quota) before your pipeline hits the wall. Works against any HTTP endpoint via generic_http; 35+ vendors pre-wired.
- Probe type 2 (scraper output quality): watches third-party Actors your pipeline depends on (e.g. TikTok, LinkedIn, Google Maps scrapers) for output schema changes, volume drops, staleness, and cost spikes.
- Alerting: exactly one deduplicated alert when a connector starts failing, and one recovery notice when it is healthy again. No alert spam.
- Pricing: $0.02 per check, billed per connector checked. No subscription.
- Agent-first: runs headlessly as an Apify Actor with JSON input and structured JSON output, so autonomous agents can schedule checks and consume results without a human in the loop.
- x402 (live): an agent can buy a single health check for $0.02 in USDC on Base at POST https://datasourcepulse.com/api/x402/check · no account of any kind. Unpaid calls return the standard 402 envelope with price and schemas. Listed in Coinbase's x402 Bazaar; payments verifiable at https://www.x402scan.com/recipient/0x7cCb91a3ad4c45097bDA4C3C29d2Cc89fA93Ab36
- x402 origin manifest: https://datasourcepulse.com/.well-known/x402.json (DiscoveredResource envelope with full input/output schemas; same content at https://datasourcepulse.com/.well-known/x402) · fetch it to learn what is payable on this origin without a facilitator or a prior transaction.

## Pages

- https://datasourcepulse.com/ · What Datasource Pulse monitors, how alerting works, supported connectors, pricing (updated 2026-08-01)
- https://datasourcepulse.com/blog/ · Blog index: field notes including the x402 agent payment story and the staged n8n endpoint failure drill (updated 2026-08-01)
- https://datasourcepulse.com/blog/an-ai-agent-just-paid-us/ · How an AI agent paid Datasource Pulse $0.02 via x402: what happened, how x402 works, on-chain proof, how agents can pay per health check (updated 2026-07-24)
- https://datasourcepulse.com/blog/we-broke-an-endpoint-on-purpose/ · Blog: a staged endpoint failure caught by the n8n + Datasource Pulse + Slack pipeline in about a minute; three importable workflow templates, monthly cost math, setup pitfalls, and why every check is JSON you design: any endpoint, key, or actor, with 35+ pre-wired vendors as a head start (updated 2026-07-25)
- https://datasourcepulse.com/blog/detect-when-an-api-key-expires-or-silently-fails/ · How to detect when an API key expires or silently fails: the scheduled canary-call pattern with response classification (401/403 dead key, empty usage counters as exhausted budget, 429 disambiguated by body), real endpoints and field names for GitHub /rate_limit with the github-authentication-token-expiration header, SerpAPI account fields plan_searches_left and extra_credits, and OpenAI /v1/models with insufficient_quota; automated route via credential checks with dead_token, quota_exhausted, rate_limited, and unexpected_response verdicts at $0.02 per check, alerts on state transitions with recovery notices (updated 2026-08-01)
- https://datasourcepulse.com/blog/alert-when-data-feed-goes-stale/ · How to get an alert when a data feed goes stale: the cron-plus-timestamp manual pattern (UTC comparison, threshold past the update cadence, empty feed as its own alarm), and supply-side freshness detection via actor checks with zero_yield streaks, volume_drop baselines, and actor_stale thresholds at $0.02 per check; generic_http covers plain HTTP feeds' liveness while timestamp math stays client-side (updated 2026-08-01)
- https://datasourcepulse.com/blog/monitor-github-personal-access-token-expiration/ · How to monitor GitHub personal access token expiration: read the github-authentication-token-expiration response header from the free /rate_limit endpoint for advance warning, the 401 vs 403 vs rate-limit disambiguation trap, honest TokenTimer comparison, and the automated daily liveness check via the pre-wired GitHub connector with dead_token and rate_limited verdicts at $0.02 per check (updated 2026-08-01)
- https://datasourcepulse.com/blog/alert-before-serpapi-quota-runs-out/ · How to get alerted before your SerpAPI quota runs out: the manual route via SerpAPI's account endpoint (real field names, the missing total_searches_used trap), threshold and budget-line advice, and the automated route via the pre-wired SerpAPI connector with a heads-up alert at 80%, a warning state at 100% of your quotaSoftCap budget line, and quota_exhausted on empty, at $0.02 per check (updated 2026-08-01)
- https://datasourcepulse.com/agents/ · Agent access: copy-paste MCP config for Apify's official server, curl and LangChain examples, the x402 pay-per-check endpoint that needs no account, and links to the n8n, Zapier, and Make automation-platform routes (updated 2026-08-01)
- https://datasourcepulse.com/integrations/n8n/ · Three importable n8n workflow templates: scraper health to Slack, daily API key + quota watchdog to Gmail, actor output drift to email; setup steps, per-month costs, verdict schema (updated 2026-07-25)
- https://datasourcepulse.com/integrations/zapier/ · Three guided Zapier templates (scraper failures to Slack, dead API key + quota watchdog to Gmail, actor output drift to email) plus two manual patterns: instant catch-hook webhooks fed by Datasource Pulse state transitions, or scheduled checks through Apify's Zapier app; input JSON included (updated 2026-07-27)
- https://datasourcepulse.com/integrations/make/ · Two Make patterns: instant custom-webhook trigger fed by Datasource Pulse state transitions, or scheduled scenario with Apify's Run an Actor module; input JSON and filter setup (updated 2026-07-24)
- https://datasourcepulse.com/blog/detect-scraper-output-schema-changes/ · How to detect scraper output schema changes and schema drift: fingerprint a known-good run's fill rates and diff each new run, three traps (spread sampling vs top-of-file, sparse-field false alarms from small-sample variance, baselines invalid across input changes), two severities (required-field break vs hollow-field warning), automated schema_drift check at $0.02 (updated 2026-08-02)
- https://datasourcepulse.com/blog/two-scrapers-changed-their-output-schema-drift-case-study/ · Schema drift case study: two Apify Store scrapers (Reddit, Instagram) silently dropped output fields on the same week, both runs exited green; drift confirmed across two consecutive daily schema_drift checks, triaged same day by asking which lost fields the consuming pipeline actually reads (none that mattered), closed by updating baselines with zero code changes; severity is defined by the consumer, not the vendor; three endings for a missing field (renamed, gated behind an input option, gone) (updated 2026-08-06)
- https://datasourcepulse.com/blog/monitor-an-apify-actor-for-failures/ · How to monitor an Apify Actor for failures, including Store actors you did not build: six alarms (actor_failed, zero_yield, schema_drift, volume_drop, cost_spike, actor_stale), why a green SUCCEEDED status misses silent failures, the input JSON, read-only token scope, $0.02 per check, and three ways to run it (updated 2026-08-02)

Section map for the home page:
- For the problem it solves → https://datasourcepulse.com/#problem
- For how checks work → https://datasourcepulse.com/#how-it-works
- For which credentials and connectors it can watch → https://datasourcepulse.com/#integrations
- For pricing → https://datasourcepulse.com/#pricing

## Run it

- Datasource Pulse on Apify: the current store listing link (install, run, JSON input/output reference, scheduling) is always at https://datasourcepulse.com/agents/
- MCP: the current config for Apify's official MCP server scoped to this Actor is always at https://datasourcepulse.com/agents/ (needs an Apify token; OAuth-capable clients can sign in instead)
- Full page text in one file: https://datasourcepulse.com/llms-full.txt
- Crawl policy: https://datasourcepulse.com/robots.txt · Sitemap: https://datasourcepulse.com/sitemap.xml