octri (Ruby)

Error and performance monitoring for Ruby backends. Report errors out of Rack or Rails with original-source context per stack frame, time every request and any sub-span you open into a waterfall, and join each server error to the client SDK error for the same request through the W3C traceparent header. In the dashboard you see the full client and server stack under one trace.

Octri turns an OpenAPI spec into a documentation site, client SDKs for ten languages, an MCP server your AI assistant can call, and monitoring for the API behind them. This gem is the Ruby monitoring runtime, and it works on its own: a generated Octri API SDK is not required. See octri.dev/monitoring.

Ruby 2.7 or newer. Standard library only. The Ruby sibling of @octri/node.

Install

gem install octri

Setup (Rack / Rails)

require "octri"
require "octri/rack"

Octri.init(
  url: "https://monitoring.example.com",
  token: ENV["OCTRI_TOKEN"],
  environment: "<your project id>",
  release: ENV["GIT_SHA"] # optional
)

# config.ru / Rails middleware stack:
use Octri::Rack

Hosted users can copy the project-scoped URL, token, and environment from the Monitoring connection settings (or its API). Omit token: only when pointing at an open self-hosted ingest endpoint. Every request carries an idempotency key.

Standalone events

No generated API SDK is required to send your own events:

Octri.capture_event(
  "checkout.completed",
  user: { id: customer.id },
  tags: { region: "eu-west", plan: "growth" },
  context: { order_id: order.id, total: order.total }
)

Delivery is best-effort and runs on a background thread. Pass event_id: to make a retried delivery idempotent.

Octri::Rack captures any exception raised by the app (linked to the request's trace) and re-raises, and times the request as a server span.

Automatic instrumentation

Octri.auto_instrument                            # traces outbound Net::HTTP calls
Octri.instrument(PG::Connection, [:exec], op: "db")  # your DB client / util class, once
Octri.instrument(cache, [:get, :set], op: "cache")

Every instrumented call (and every outbound HTTP request) becomes a sub-span under the current request, with no per-call code. Calls to your monitoring backend are never traced (no feedback loop).

Sub-spans (where time goes)

Octri.span("orders.list", op: "db") do
  Order.where(status: "open").to_a
end

# or manual control:
s = Octri.start_span("cache.get", op: "cache")
value = cache.read(key)
s.finish

op ("db", "cache", "http", and so on) color-codes the waterfall; nested Octri.span calls nest correctly.

Manual error capture

begin
  risky!
rescue => e
  Octri.capture_error(e)
  raise
end

---

What gets redacted

Payloads are scrubbed on the way out, so a credential that ended up in a log line or a context object never reaches the dashboard.

Any key whose name looks like a credential (password, secret, token, apiKey, authorization, cookie, ssn and the rest of the usual list) has its value replaced with [redacted], at any depth. Matching ignores case and separators, so api_key, apiKey and X-API-KEY are all the same key.

Free text is swept too: the message, an error message and its stack, and anything else you send as a string. Bearer tokens, JWTs, card numbers and email addresses come out as [redacted]. A card number has to pass the Luhn check first, so an order number or a timestamp survives.

user is the exception. It is the field you fill with an identity on purpose, so user.email is reported exactly as you set it. Credential-shaped keys inside it are still redacted.

Add your own key names:

Octri.add_scrub_fields("account_number", "otp")

Or take the payload yourself, and return nil to drop the event:

Octri.set_before_send { |payload| payload[:path] == "/health" ? nil : payload }

Redaction runs after your hook, so a hook cannot leak a credential by accident.

The rest of Octri

Product	What it does
API Studio	Your OpenAPI spec becomes a hosted documentation site with a live request playground, editable page by page.
SDK Studio	The same spec becomes client libraries for ten languages, versioned and released together.
MCP	Your endpoints and docs become tools an AI assistant can call, generated from the same spec.
Monitoring	Errors, traces, uptime and releases for the API, joined to the SDK calls that reached it.

Monitoring runtimes

• Node
• Python
• Go
• Ruby
• Rust
• PHP
• Java
• Kotlin
• Swift
• Dart

More

• Documentation
• Pricing
• Changelog

MIT licensed.