Coolhand Ruby Monitor

Monitor and log LLM API calls — OpenAI, Anthropic, Google Gemini, Cohere, and any Faraday-based or Net::HTTP client — to the Coolhand analytics platform. Supports Ruby LLM monitoring, request logging, and feedback collection.

Scope: Coolhand intercepts outgoing HTTP requests to configured LLM API endpoints only. It does not read, scan, or transmit your source code files.

gem 'coolhand'

1. Get API Key: Visit coolhandlabs.com to create a free account
2. Install: gem install coolhand
3. Initialize: Add configuration to your Ruby application
4. Configure: Set your API key in the configuration block
5. Deploy: Your AI calls are now automatically monitored!

🔥 Set it and forget it! Monitor ALL AI API calls across your entire application with minimal configuration.

# Add this configuration at the start of your application
require 'coolhand'

Coolhand.configure do |config|
  config.api_key = 'your_api_key_here'
  config.silent = true  # Set to false for debug output
end

# That's it! ALL AI API calls are now automatically monitored:
# ✅ OpenAI SDK calls
# ✅ Anthropic API calls
# ✅ Direct HTTP requests to AI APIs
# ✅ Any library making AI API calls via Faraday's default adapter

# NO code changes needed in your existing services!

✨ Why Automatic Monitoring:

• 🚫 Zero refactoring - No code changes to existing services
• 📊 Complete coverage - Monitors AI libraries using Net::HTTP or Faraday's default adapter automatically
• 🔒 Security built-in - Automatic credential sanitization
• ⚡ Performance optimized - Negligible overhead via async logging
• 🛡️ Future-proof - Automatically captures new AI calls added by your team

Point the SDK at your own Coolhand-compatible endpoint via config.base_url for compliance or data-residency requirements. See Self-Hosted Deployments →.

Collect feedback on LLM responses to improve model performance.

Frontend Feedback Widget: For browser-based feedback collection, see coolhand-js - an accessible, lightweight JavaScript widget that leverages best UX practices to capture actionable user feedback on any AI output.

require 'coolhand'

# Create feedback for an LLM response
feedback_service = Coolhand::FeedbackService.new

feedback = feedback_service.create_feedback(
  llm_request_log_id: 'abc123def456', # hashid from a prior response; a raw integer FK also still works
  llm_provider_unique_id: 'req_xxxxxxx',
  client_unique_id: 'workorder-chat-456',
  creator_unique_id: 'user-789',
  original_output: 'Here is the original LLM response!',
  revised_output: 'Here is the human edit of the original LLM response.',
  explanation: 'Tone of the original response read like AI-generated open source README docs',
  sentiment: 'dislike'
)

For a full field reference and matching strategy, see Feedback API →.

Read back the LLM request templates your logs are matched against. Unlike the logging and feedback methods, these require your private API key — the public key is write-only on this API and is rejected exactly like an invalid one.

require 'coolhand'

templates = Coolhand.template_service

result = templates.search_templates(search: 'summar', status: 'published')
result.templates.each { |template| puts "#{template[:name]} (#{template[:log_count]} logs)" }
result.pagination.total_count

# Prompt patterns come from a single-template fetch only
detail = templates.get_template(result.templates.first[:id])
detail[:user_prompt_pattern]

Search is a parameter on the list endpoint, not a route of its own, and the Unmatched / Ignored API Calls system buckets are hidden unless you pass include_system: true. These are read methods, so they raise Coolhand::HttpError where the write methods log and return nil.

For the full filter reference, pagination, and error handling, see Reading Templates →.

Pass include_metrics: true to search_templates (or search_workloads) for the dashboard's cost and performance numbers, over a rolling days_back window or an explicit since: / until: one. search_logs carries each log's cost and can sort and filter by it.

window = { since: Time.utc(2026, 9, 1), until: Time.utc(2026, 10, 1) }

Coolhand.workload_service.search_workloads(include_metrics: true, **window).workloads
  .map { |w| [w[:name], w[:metrics][:total_cost]] }

Coolhand.log_service.search_logs(order: 'cost_desc', min_cost: 1.0, per: 10, **window).logs

See Reading Workloads → and Reading Logs →.

Coolhand.help_center_service reads the Coolhand help center (the same pages as coolhandlabs.com/help and /updates) with no API key required:

require 'coolhand'

help_center = Coolhand.help_center_service

result = help_center.search_help_articles(query: 'match rate', per_page: 5)
article = help_center.get_help_article(result.articles.first[:slug]) # full markdown :body
categories = help_center.list_help_categories

Anonymous callers are limited to 30 requests per minute per IP. If config.api_key is set it is sent and lifts the limit; a key the server rejects raises a 401 Coolhand::HttpError. These are read methods, so they raise where the write methods log and return nil.

See Reading the Help Center → for parameters, the category filter and errors.

Attach feedback to an optimization as evidence, one at a time or in bulk. These require your private API key and raise Coolhand::Error on failure (Coolhand::HttpError for non-2xx responses).

links = Coolhand.optimization_feedback_link_service

link = links.link_feedback('optimizationHashid', 'feedbackHashid', note: 'why')
links.bulk_link_feedback('optimizationHashid', %w[fb1 fb2 fb3]) # => { linked:, already_linked:, errored:, not_found: }
links.unlink_feedback('optimizationHashid', link[:id])

For batching, error semantics, and details, see Linking Feedback →.

Tell Coolhand which workload and template an LLM call belongs to, instead of waiting for it to be matched. Calls captured inside the block carry the declared identity:

Coolhand.with_context(reporter: 'my-app', workload: { key: 'Parser#call' }, template: { key: 'Parser#call/summarize' }) do
  client.chat(...) # captured as usual
end

Coolhand.workload_service.upsert(key: 'Parser#call', name: 'Parser') # private key; create or rename

Or declare it once on the class and let every capture inside it pick it up, subclasses included:

class ApplicationService
  include Coolhand::Declarable
  coolhand_declare_all # each public method declares "ClassName#method"
end

For nesting rules, class-level declarations, Rails jobs and controllers, upserts, and public-key limits, see Declared Workloads →.

Create an initializer file at config/initializers/coolhand.rb:

# config/initializers/coolhand.rb
Coolhand.configure do |config|
  # Your Coolhand API Key (Required)
  # Best practice: Use Rails credentials or environment-specific configuration
  config.api_key = Rails.application.credentials.coolhand_api_key

  # Set to true to suppress console output
  config.silent = Rails.env.production?

  # Specify which LLM endpoints to intercept (array of strings)
  # Optional - defaults to every provider listed under "Supported Libraries" below
  # config.intercept_addresses = ["api.openai.com", "api.anthropic.com", "api.cohere.com/v2/chat"]
end

class ChatController < ApplicationController
  def create_feedback
    feedback_service = Coolhand::FeedbackService.new

    feedback = feedback_service.create_feedback(
      llm_request_log_id: params[:log_id],
      creator_unique_id: current_user.id,
      original_output: params[:original_response],
      revised_output: params[:edited_response],
      explanation: params[:feedback_text],
      sentiment: params[:sentiment]
    )

    if feedback
      render json: { success: true, message: 'Feedback recorded' }
    else
      render json: { success: false, message: 'Failed to record feedback' }, status: 422
    end
  end
end

class FeedbackCollectionJob < ApplicationJob
  def perform(feedback_data)
    feedback_service = Coolhand::FeedbackService.new

    feedback_service.create_feedback(
      llm_provider_unique_id: feedback_data[:request_id],
      creator_unique_id: feedback_data[:user_id],
      original_output: feedback_data[:original],
      explanation: feedback_data[:reason],
      sentiment: feedback_data[:sentiment]
    )
  end
end

require 'openai'
require 'coolhand'

# Configure Coolhand
Coolhand.configure do |config|
  config.api_key = 'your_api_key_here'
end

# Use OpenAI normally - requests are automatically logged
client = OpenAI::Client.new(access_token: ENV['OPENAI_API_KEY'])

response = client.chat(
  parameters: {
    model: "gpt-4o-mini",
    messages: [{ role: "user", content: "These pretzels are making me thirsty!"}],
    temperature: 0.7
  }
)

puts response.dig("choices", 0, "message", "content")
# The request and response have been automatically logged to Coolhand!

📖 Complete Anthropic Integration Guide → - Supports both official and community gems with automatic detection

Use Coolhand.with_capture and Coolhand.without_capture to override the global capture setting for a specific block of code, without changing your configuration.

Capture a specific call when capture is globally disabled:

Coolhand.configure do |config|
  config.api_key = 'your_api_key_here'
  config.capture = false  # disabled globally
end

Coolhand.with_capture do
  response = openai_client.chat(parameters: {
    model: "gpt-4o-mini",
    messages: [{ role: "user", content: "Hello!" }],
  })
  # This request is captured and forwarded to Coolhand
end

Skip capture for a specific call when capture is globally enabled:

Coolhand.without_capture do
  response = openai_client.chat(parameters: { ... })
  # This request is NOT forwarded to Coolhand
end

Both methods are thread-safe. Blocks nest correctly — the innermost block takes precedence, and the outer setting is restored when the block exits. The override is fiber-local, so it does not follow into new Threads, Fibers or Async tasks started inside the block; see Threads and background logging.

For inbound webhooks (like audio transcripts or tool calls), the automatic interceptor won't capture them since they're incoming requests TO your application. In these cases, use the simple forward_webhook helper method:

class WebhooksController < ApplicationController
  def elevenlabs
    raw_body = request.body.read
    webhook_data = JSON.parse(raw_body)

    # Forward to Coolhand with automatic field generation and binary filtering
    Thread.new do
      Coolhand.logger_service.forward_webhook(
        webhook_body: webhook_data,        # Required: webhook payload
        source: "elevenlabs",              # Required: service name
        event_type: webhook_data["type"],  # Optional: e.g., post_call_transcription
        headers: request.headers,          # Optional & recommended
      )
    end

    render json: { status: "success" }
  end
end

Required Parameters:

• webhook_body - The webhook payload (Hash or parsed JSON)
• source - Service name (String, e.g., "elevenlabs", "stripe", "twilio")

Optional Parameters:

• event_type - Event type to append to URL (e.g., "post_call_transcription" → webhook://elevenlabs/post_call_transcription)
• headers - Request headers (automatically sanitized). Passing Rails request.headers is safe: when the input looks like a Rack env (it has a REQUEST_METHOD key or a lowercase dotted key such as rack.input, or every key is UPPER_CASE; in a plain hash use hyphenated or mixed-case names), only HTTP_* headers, CONTENT_TYPE and CONTENT_LENGTH are forwarded, so entries such as rack.*, REQUEST_METHOD, QUERY_STRING, or process env vars (DATABASE_URL, ...) are dropped
• url - Optional URL override; credentials in userinfo and sensitive query params are redacted, and any #fragment is dropped
• conversation_id, agent_id, metadata - Custom fields for your tracking needs

Error Handling:

• Silent mode = false: Raises ArgumentError if required parameters are missing
• Silent mode = true: Logs warning and returns false if required parameters are missing

That's it! The forward_webhook method automatically:

• ✅ Generates unique ID and timestamp
• ✅ Filters out binary data (audio, images, etc.)

The monitor captures:

• Request Data: Method, URL, headers, request body
• Response Data: Status code, headers, response body
• Metadata: Timestamp, protocol used
• LLM-Specific: Model used, token counts, temperature settings

Headers containing API keys are automatically sanitized for security.

Coolhand does not track or store binary data. The gem automatically filters out:

• Audio files and data (audio, audio_data, full_audio, raw_audio)
• Image data (image_data, image_content)
• File content (file_content, binary_content)
• Base64 encoded data (audio_base64, base64_data)
• Voice samples and audio URLs

This ensures:

• ✅ Smaller payloads - Only text and metadata are sent
• ✅ Faster processing - No bandwidth wasted on binary data
• ✅ Privacy focused - Audio/video content never leaves your infrastructure
• ✅ Clean logs - Focus on conversational data, not media files

The filtering is automatic and applies to all monitored API calls and webhook logging.

The monitor works with multiple transport layers and Ruby libraries:

Faraday-based libraries (via Faraday's default net_http adapter — a non-Net::HTTP adapter such as faraday-typhoeus is not covered):

• OpenAI Ruby SDK
• ruby-anthropic gem (community Anthropic gem)
• ruby-openai gem
• LangChain.rb
• Direct Faraday requests using the default adapter

Native HTTP libraries:

• Official Anthropic Ruby SDK (using Net::HTTP)
• GitHub Models SDK / any client using models.github.ai
• Any library using Net::HTTP directly

Other providers monitored out of the box:

• Google Gemini (generativelanguage.googleapis.com)
• Google Vertex AI (aiplatform.googleapis.com) — see the Vertex AI batch result logging guide for async batch jobs
• Azure OpenAI Service, Azure AI Foundry, and Azure Machine Learning managed online endpoints (dedicated hosts, multi-service Azure AI Services hosts, and serverless/MaaS deployments) — see Configuration for the full list of covered hosts
• AWS Bedrock (bedrock-runtime.<region>.amazonaws.com, plus the .amazonaws.com.cn, .api.aws dual-stack and bedrock-runtime-fips.<region> variants — model invoke, Converse and the OpenAI-compatible endpoint)
• Cloudflare AI Gateway
• OpenRouter
• OpenCode Zen (opencode.ai)
• DeepSeek (api.deepseek.com)
• Mistral (api.mistral.ai)
• Perplexity (api.perplexity.ai)
• xAI (api.x.ai)
• Cohere (chat and embed endpoints only, on api.cohere.com and api.cohere.ai)
• TypeSafe Jev / System One (api.typesafe.ai/v1/systemone)
• Ollama (self-hosted, default port 11434 — chat, generate and embed endpoints) — see Configuration for non-default hosts

Universal Coverage: Since most Ruby HTTP libraries use Net::HTTP under the hood, Coolhand's single interceptor provides comprehensive monitoring without needing library-specific integrations.

Coolhand uses a unified Net::HTTP interceptor to capture outgoing requests to configured LLM API endpoints:

• Extends Ruby's Net::HTTP using Module#prepend (a standard Ruby technique for wrapping library behavior)
• Covers HTTP libraries that delegate to Net::HTTP under the hood (which is most of them)
• Handles both standard requests and streaming responses via read_body interception
• Thread-safe design using thread-local storage for streaming buffers

When a request matches configured LLM endpoints:

1. The original request executes normally
2. Request and response data (body, headers, status) are captured by the interceptor
3. For streaming requests, the complete accumulated response is captured (not individual chunks)
4. Data is sent to the Coolhand API inline, after the original response is available (bounded by a 5-second connect/read timeout so a slow or unreachable Coolhand backend can't hang your request indefinitely)
5. Your application continues once the log is sent (or times out) — the original request/response is never blocked on Coolhand being reachable

For non-matching endpoints, requests pass through unchanged.

Enable verbose logging to see what's being intercepted:

Coolhand.configure do |config|
  config.api_key = 'your_api_key_here'
  config.silent = false  # Enable console output
end

In test environments, you may want to configure differently:

# config/initializers/coolhand.rb
if Rails.env.test?
  Coolhand.configure do |config|
    config.api_key = 'test_key'
    config.silent = true
  end
end

For standard Ruby scripts or non-Rails applications:

#!/usr/bin/env ruby
require 'coolhand'

Coolhand.configure do |config|
  config.api_key = 'your_api_key_here'  # Store securely, don't commit to git
  config.environment = 'production'
  config.silent = false
end

# Your application code here...

🆓 Sign up for free at coolhandlabs.com to get your API key and start monitoring your LLM usage.

What you get:

• Complete LLM request and response logging
• Usage analytics and insights
• Feedback collection and quality scoring
• No credit card required to start

The monitor handles errors gracefully:

• Failed API logging attempts are logged to console but don't interrupt your application
• Invalid API keys will be reported but won't crash your app
• Network issues are handled with appropriate error messages
• The read methods (search_templates, get_template, search_workloads, search_logs, get_log, search_help_articles, get_help_article, list_help_categories), the feedback-link methods and the upsert methods are the exception — they raise Coolhand::Error (Coolhand::HttpError for non-2xx responses); see Reading Templates →, Reading Workloads →, Reading Logs →, Linking Feedback → and Declared Workloads →

• Configuration — Self-hosted deployments, base_url rules, debug mode, custom intercept addresses
• Feedback API — Full field reference, matching strategies, sentiment values
• Reading Templates — Search LLM request templates and fetch a single one, prompt patterns included, using the private API key
• Reading Workloads — List workloads with optional cost and performance metrics over a rolling or explicit time window
• Reading Logs — Search logs with per-log cost, filter and sort by cost, and fetch one log with its cost breakdown
• Reading the Help Center — Search and read Coolhand's help center articles and list its sections, with no API key required
• Linking Feedback — Link feedback to an optimization as evidence, singly or in bulk, using the private API key
• Declared Workloads — Declare the workload and template of captured LLM calls with Coolhand.with_context, and create or rename them with upsert
• Anthropic Integration — Official and community Anthropic Ruby gems, streaming, using both gems, and troubleshooting
• ElevenLabs Integration — Webhook capture, feedback submission, and Rails integration
• OpenAI Batch Webhook Handler — Handle OpenAI batch job completion events via webhook interception
• Google Vertex AI Batch Result Logging — Log completed Vertex AI batch prediction job results
• Examples — Runnable scripts that make a real provider call through Coolhand, useful as a live smoke test

• API keys in request headers and credential-like URL query params are automatically redacted (provider rate-limit headers such as x-ratelimit-remaining-tokens are kept)
• Credentials in request and response bodies (MCP authorization/authorization_token, Realtime ephemeral client secrets, api_key, and similar) are redacted; see Body redaction
• No sensitive data is exposed in logs
• All data is sent via HTTPS to Coolhand servers

• Frontend (Feedback Collection Widget): coolhand-js - Frontend feedback widget for collecting user feedback on AI outputs
• Node.js: coolhand-node package - Coolhand monitoring for Node.js applications
• Python: coolhand package - Coolhand monitoring for Python applications

• Questions? Create an issue
• Contribute? Submit a pull request
• Support? Visit coolhandlabs.com

Coolhand Labs builds LLM observability and feedback tooling so teams can monitor, understand, and improve their AI applications. Learn more at coolhandlabs.com.

Apache-2.0