Get started with Apidepth

Instrument your app once and get latency percentiles, error rates, rate limit tracking, and fleet benchmarks across every vendor API you call — measured from your server, not a probe running somewhere else.

Prerequisites

  • An Apidepth API key — get one from the API keys page

Install the SDK

Choose your language. Each SDK auto-instruments your HTTP layer — no changes to existing request code. Follow the full setup guide on GitHub for configuration options, framework integrations, and troubleshooting.

RubyRubyGems

Gemfile

gem "apidepth"

Configure

Apidepth.configure do |c|
  c.api_key = ENV["APIDEPTH_API_KEY"]
end
Full setup guide on GitHub →
PythonPyPI

pip

pip install apidepth

Configure

apidepth.configure(
    api_key=os.environ["APIDEPTH_API_KEY"],
)
apidepth.instrument()
Full setup guide on GitHub →
JavaScript / TypeScriptnpm

npm

npm install apidepth

Configure

Apidepth.configure({
  apiKey: process.env.APIDEPTH_API_KEY,
});
Apidepth.instrument();
Full setup guide on GitHub →
Tip:Free tier: Track up to 3 vendors at no cost — no credit card, no expiry. Upgrade to Solo when you need more vendors or 90-day retention.

Guided setup (optional)

Each SDK ships a setup subcommand that walks you through configuration interactively — detects your framework, generates the correct initializer snippet, and optionally writes it to disk:

# Ruby
bundle exec apidepth setup

# Python
python -m apidepth setup

# JavaScript / TypeScript
npx apidepth setup

For CI/CD or AI-assisted environments, skip all prompts with flags:

# Ruby
bundle exec apidepth setup --api-key $APIDEPTH_API_KEY --no-prompt

# Python
python -m apidepth setup --api-key $APIDEPTH_API_KEY --no-prompt

# JavaScript / TypeScript
npx apidepth setup --api-key $APIDEPTH_API_KEY --no-prompt

See the CLI reference for the full flag list including --framework, --ignored-hosts, and --collector-url.

Confirm connectivity

After installing, send a synthetic test event to confirm the SDK is wired up and can reach the collector. This event is never counted in your vendor data or alerts — it only triggers the dashboard confirmation banner.

# Ruby
bundle exec apidepth test

# Python
python -m apidepth test

# JavaScript / TypeScript
npx apidepth test

On success you'll see ✓ Test event received by collector. The Overview page will show a green confirmation banner until your first real production event arrives.

Verify with real traffic

Make any outbound request to a supported vendor from your running app — Stripe, OpenAI, Anthropic, GitHub, Twilio, SendGrid, HubSpot, Datadog, and many more. Open the Overview page — your first event should appear within 30 seconds.

What gets captured

Every outbound HTTP request to a recognised vendor produces one event. All fields below are captured identically across every SDK.

FieldDescription
vendorVendor slug, e.g. "stripe", "openai"
endpointNormalized path, e.g. "/v1/charges/:id"
methodHTTP verb: "GET", "POST", etc.
statusHTTP status code, or null on timeout
outcome"success", "client_error", "server_error", "timeout", "unknown"
duration_msWall-clock time in milliseconds, including DNS and TLS on first connection
cold_starttrue if this request paid for the TLS handshake; excluded from p95 calculations
envEnvironment tag from your SDK config
tsUnix timestamp in milliseconds
rl_remainingRemaining quota — present when vendor rate limit headers are found
rl_limitTotal quota — present when vendor rate limit headers are found
rl_reset_atQuota reset time in epoch milliseconds — present when vendor rate limit headers are found

What is never captured

  • Request or response bodies
  • Request or response headers (including Authorization)
  • Query string parameters
  • Any credential, token, or secret your application uses to authenticate with a vendor
  • User identifiers or PII of any kind

Path normalization strips resource IDs before the event leaves your server. /v1/charges/ch_3Ox4Kz2e becomes /v1/charges/:id.

Features

Latency monitoring

p50, p95, and p99 latency for every vendor endpoint, aggregated hourly. Measured from your server — the number your users actually feel.

Fleet benchmarking

Your latency percentiles are compared against anonymized fleet data from all Apidepth customers. See whether a slow Stripe response is isolated to your infrastructure or industry-wide.

Proof of Innocence

When all endpoints to a vendor spike together, Apidepth surfaces a verdict: isolated (your code or infra) or tracking (vendor-side). Tell ops "it's Stripe, not us" in one card.

Rate Limit Intelligence →

Tracks 429 patterns and captures quota headers from every response to project burn-down before you hit the ceiling.

Vendor status →

TCP/TLS probes and status-page polling every 60 seconds across all built-in vendors. Always current, independent of your SDK traffic.

Alerts & weekly digest →

Alerts fire on latency spikes, error rate thresholds, 429s, and quota burn-down. Delivered via email, Slack, or outbound webhook.

Next steps