Observability

Feature Overview: Observability is iFlow CLI's monitoring and analysis system, providing performance monitoring, usage analysis, and debugging support.

Learning Time: 10-15 minutes

Prerequisites: Understanding basic JSON configuration, familiarity with system monitoring concepts

What is Observability

Observability is a monitoring and analysis system provided by iFlow CLI that enables you to gain deep insights into CLI performance, operational status, and usage patterns. By enabling observability features, you can obtain tracing data, performance metrics, and structured logs to better monitor operations, debug issues, and optimize user experience.

Core Features

Feature	Description	Advantage
Standardized Protocol	Built on OpenTelemetry standards	Compatible with various monitoring backends
Multiple Data Types	Supports traces, metrics, and logs	Comprehensive observability
Flexible Configuration	Supports local and cloud output methods	Adapts to different deployment environments
Privacy Protection	Configurable sensitive information recording	Protects user privacy
Performance Optimization	Asynchronous processing, doesn't affect CLI performance	Transparent monitoring

How It Works

Observability Data Flow

CLI Operations → Data Collection → Data Processing → Data Output → Monitoring Analysis
    ↓
[User Behavior] → [Trace Recording] → [Metric Aggregation] → [Local/Cloud] → [Performance Insights]

Technical Architecture

• Data Collection: Automatically collects performance data based on OpenTelemetry SDK
• Data Processing: Structured processing with filtering and aggregation support
• Data Output: Supports multiple outputs including local files, OTLP endpoints, cloud services
• Privacy Protection: Configurable data masking and filtering mechanisms

Detailed Feature Description

Configuration Management

Observability features support multiple flexible configuration methods, primarily managed through configuration files and environment variables, with CLI flags able to override specific session settings.

Configuration Priority

Priority	Configuration Method	Description
Highest	CLI flags	Temporarily override current session
High	Environment variables	Global environment configuration
Medium	Project configuration	.iflow/settings.json
Low	User configuration	~/.iflow/settings.json
Lowest	Default values	System default settings

CLI Flag Parameters

Parameter	Function	Example
--telemetry / --no-telemetry	Enable/disable observability	iflow --telemetry
--telemetry-target <local|gcp>	Set output target	iflow --telemetry-target local
--telemetry-otlp-endpoint <URL>	Set OTLP endpoint	iflow --telemetry-otlp-endpoint http://localhost:4317
--telemetry-outfile <path>	Export to file	iflow --telemetry-outfile ./metrics.json
--telemetry-log-prompts	Record prompts	iflow --telemetry-log-prompts

Environment Variables

Variable Name	Purpose	Example
OTEL_EXPORTER_OTLP_ENDPOINT	Set OTLP export endpoint	http://localhost:4317

Default Settings

Configuration Item	Default Value	Description
telemetry.enabled	false	Observability disabled by default
telemetry.target	local	Local output
telemetry.otlpEndpoint	http://localhost:4317	Local OTLP endpoint
telemetry.logPrompts	true	Record prompts

Configuration Examples

Local Development Environment

Enable local observability in .iflow/settings.json:

{
  "telemetry": {
    "enabled": true,
    "target": "gcp"
  },
  "sandbox": false
}

Export to File

You can export all observability data to local files for detailed analysis and inspection.

Simply use the --telemetry-outfile flag and specify the output file path to enable file export functionality. Note that this feature needs to be used with --telemetry-target=local.

iflow --telemetry --telemetry-target=local --telemetry-outfile=/path/to/telemetry.log "your prompt"

Running OTEL Collector

The OTEL collector is a powerful service responsible for receiving, processing, and exporting observability data. The CLI sends data through the efficient OTLP/gRPC protocol.

Want to learn more about standard OTEL exporter configurations? Check the official documentation for detailed information.

Local Deployment

Using the npm run telemetry -- --target=local command can easily automate the setup of local observability pipelines, including automatically configuring necessary settings in the .iflow/settings.json file. The script will help you install otelcol-contrib (OpenTelemetry Collector) and jaeger (Jaeger UI for visualizing traces).

Usage Steps:

1. Run Command: Execute the following command in the repository root:

   npm run telemetry -- --target=local

   The script will automatically complete the following work for you:

   • Download Jaeger and OTEL components as needed
   • Start local Jaeger instance
   • Start configured OTEL collector to receive iFlow CLI data
   • Automatically enable observability in your workspace settings
   • Automatically disable observability when exiting

2. View Traces: Open your browser and visit http://localhost:16686 to use Jaeger UI. Here you can view detailed trace information for iFlow CLI operations in depth.

3. Check Logs and Metrics: The script will save OTEL collector output (including logs and metrics) to ~/.iflow/tmp/<projectHash>/otel/collector.log. The script provides convenient viewing links and local commands to track your observability data.

4. Stop Services: Press Ctrl+C in the terminal running the script to stop the OTEL collector and Jaeger services.

Logs and Metrics Reference

The following sections detail the structure of logs and metrics generated by iFlow CLI to help you better understand and analyze the data.

• All logs and metrics include sessionId as a common identifying attribute.

Logs

Logs record timestamped information for specific events. iFlow CLI records the following important events:

• iflow_cli.config: This event occurs once at startup and contains CLI configuration.

  • Attributes:
    • model (string)
    • embedding_model (string)
    • sandbox_enabled (boolean)
    • core_tools_enabled (string)
    • approval_mode (string)
    • api_key_enabled (boolean)
    • vertex_ai_enabled (boolean)
    • code_assist_enabled (boolean)
    • log_prompts_enabled (boolean)
    • file_filtering_respect_git_ignore (boolean)
    • debug_mode (boolean)
    • mcp_servers (string)

• iflow_cli.user_prompt: This event occurs when user submits a prompt.

  • Attributes:
    • prompt_length
    • prompt (excluded if log_prompts_enabled configuration is false)
    • auth_type

• iflow_cli.tool_call: This event occurs for each function call.

  • Attributes:
    • function_name
    • function_args
    • duration_ms
    • success (boolean)
    • decision (string: "accept", "reject", or "modify", if applicable)
    • error (if applicable)
    • error_type (if applicable)

• iflow_cli.api_request: This event occurs when making requests to iFlow API.

  • Attributes:
    • model
    • request_text (if applicable)

• iflow_cli.api_error: This event occurs when API requests fail.

  • Attributes:
    • model
    • error
    • error_type
    • status_code
    • duration_ms
    • auth_type

• iflow_cli.api_response: This event occurs when receiving responses from iFlow API.

  • Attributes:
    • model
    • status_code
    • duration_ms
    • error (optional)
    • input_token_count
    • output_token_count
    • cached_content_token_count
    • thoughts_token_count
    • tool_token_count
    • response_text (if applicable)
    • auth_type

• iflow_cli.flash_fallback: This event occurs when iFlow CLI switches to flash as fallback.

  • Attributes:
    • auth_type

• iflow_cli.slash_command: This event occurs when user executes slash commands.

  • Attributes:
    • command (string)
    • subcommand (string, if applicable)

Metrics

Metrics provide numerical measurements of behavior over time. iFlow CLI collects the following key metrics:

• iflow_cli.session.count (counter, integer): Increments once each time CLI starts.

• iflow_cli.tool.call.count (counter, integer): Counts tool calls.

  • Attributes:
    • function_name
    • success (boolean)
    • decision (string: "accept", "reject", or "modify", if applicable)

• iflow_cli.tool.call.latency (histogram, milliseconds): Measures tool call latency.

  • Attributes:
    • function_name
    • decision (string: "accept", "reject", or "modify", if applicable)

• iflow_cli.api.request.count (counter, integer): Counts all API requests.

  • Attributes:
    • model
    • status_code
    • error_type (if applicable)

• iflow_cli.api.request.latency (histogram, milliseconds): Measures API request latency.

  • Attributes:
    • model

• iflow_cli.token.usage (counter, integer): Counts token usage.

  • Attributes:
    • model
    • type (string: "input", "output", "thought", "cache", or "tool")

• iflow_cli.file.operation.count (counter, integer): Counts file operations.

  • Attributes:
    • operation (string: "create", "read", "update"): Type of file operation.
    • lines (integer, if applicable): Number of lines in file.
    • mimetype (string, if applicable): File MIME type.
    • extension (string, if applicable): File extension.

Cloud Deployment Environment

For cloud deployment, refer to the following platform observability integration guides:

Google Cloud Platform (GCP)

Configuration Example

{
  "telemetry": {
    "enabled": true,
    "target": "gcp",
    "gcpProjectId": "your-project-id",
    "gcpServiceAccount": "path/to/service-account.json"
  }
}

Environment Variable Setup

export GOOGLE_APPLICATION_CREDENTIALS="path/to/service-account.json"
export GOOGLE_CLOUD_PROJECT="your-project-id"

Other Cloud Platforms

AWS Integration

• Supports sending trace data to AWS X-Ray via OTLP protocol
• Configurable CloudWatch metrics collection

Azure Integration

• Supports Application Insights integration
• Provides Azure Monitor compatible data formats

Usage Examples

Basic Configuration Enablement

Enable Observability at CLI Startup

# Enable observability and set local output
iflow --telemetry --telemetry-target local

# Enable observability and set OTLP endpoint
iflow --telemetry --telemetry-otlp-endpoint http://localhost:4317

# Enable observability and log prompt content
iflow --telemetry --telemetry-log-prompts

Project-level Configuration

Create .iflow/settings.json in project root:

{
  "telemetry": {
    "enabled": true,
    "target": "local",
    "otlpEndpoint": "http://localhost:4317",
    "logPrompts": true,
    "outputFile": "./logs/telemetry.json"
  }
}

Global User Configuration

Configure in ~/.iflow/settings.json:

{
  "telemetry": {
    "enabled": false,
    "target": "local",
    "defaultEndpoint": "http://localhost:4317"
  }
}

Advanced Use Cases

Performance Monitoring

# Enable detailed performance monitoring
iflow --telemetry --telemetry-target local --telemetry-outfile performance.log

# Execute task and monitor performance
"Please analyze this large project"

# View performance data
cat performance.log | jq '.metrics[] | select(.name == "iflow_cli.api.request.latency")'

Team Collaboration Monitoring

# Team shared OTLP collector
iflow --telemetry --telemetry-otlp-endpoint http://team-collector:4317

# Set team identifier
export OTEL_RESOURCE_ATTRIBUTES="team=frontend,project=webapp"

Debug Mode Data Collection

# Enable complete debug observability
iflow --telemetry --telemetry-log-prompts --telemetry-outfile debug.log

# Execute problematic operation
"Execute potentially problematic operation"

# Analyze debug data
grep "error" debug.log | jq .

Troubleshooting

Common Issues and Solutions

Issue	Possible Cause	Solution
Observability data not recorded	Observability feature not enabled	Use --telemetry flag or enable via configuration file
OTLP connection failed	Endpoint not accessible or configuration error	Check endpoint URL and network connection
File export failed	Insufficient disk space or permission issues	Check disk space and file write permissions
Incomplete data	Filtering configuration too strict	Check data filtering and masking settings
Noticeable performance impact	Observability data volume too large	Adjust sampling rate or reduce recording details

Diagnostic Steps

1. Basic Configuration Check

# Check current configuration status
/about

# View observability-related settings
cat .iflow/settings.json | jq '.telemetry'

# Verify if CLI flags take effect
iflow --help | grep telemetry

2. Connectivity Test

# Test OTLP endpoint connection
curl -X POST http://localhost:4317/v1/traces -H "Content-Type: application/json" -d '{}'

# Check local collector status
ps aux | grep otelcol

# Verify network port is open
netstat -ln | grep 4317

3. Data Validation

# Check output file content
tail -f /path/to/telemetry.log

# Verify data format
cat telemetry.log | jq '.traces[0]'

# Count data volume
wc -l telemetry.log

4. Performance Analysis

# Monitor CLI performance
time iflow --telemetry "Simple test command"

# Check observability overhead
iflow --telemetry --telemetry-outfile perf.log "test" && ls -la perf.log

# Compare performance with/without observability
time iflow "same command" # without observability
time iflow --telemetry "same command" # with observability

Data Privacy Protection

Sensitive Information Filtering

• Prompt Protection: Use --no-telemetry-log-prompts to disable prompt recording
• Parameter Filtering: Automatically filters tool parameters containing passwords, keys
• Path Masking: User path information is hashed
• Content Cleaning: Automatically identifies and cleans sensitive content

Data Storage Security

• Local Storage: Data stored only in local file system by default
• Transmission Encryption: OTLP protocol supports TLS encrypted transmission
• Access Control: Configuration files and log files permission control
• Data Retention: Configurable automatic data cleanup policies

Platform Compatibility

Platform	Support Level	Special Notes
Windows	Full support	Path separators automatically handled, supports PowerShell
macOS	Full support	Supports system-level OpenTelemetry integration
Linux	Full support	Native OpenTelemetry support, complete container compatibility

Containerized Deployment Considerations

• Docker Environment: Requires proper network and volume mount configuration
• Kubernetes Integration: Supports data collection through Service Mesh
• Permission Management: Requires appropriate file system permissions within containers