AI Configuration

iqtoolkit-analyzer can use AI providers to generate enhanced tuning recommendations from your PostgreSQL metrics.

Supported Providers

Provider	Flag Value	Model Examples
OpenAI	openai	gpt-4o, gpt-4o-mini
Anthropic	anthropic	claude-sonnet-4-5, claude-sonnet-4-20250514
Google Gemini	gemini	gemini-pro
Kiro (Amazon Bedrock)	kiro	Uses AWS credentials

Environment Variables

Set the API key for your chosen provider:

# OpenAI
export OPENAI_API_KEY="sk-..."

# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."

# Google Gemini
export GEMINI_API_KEY="AI..."

# Kiro (Amazon Bedrock) — uses standard AWS credentials
export AWS_REGION="us-east-1"

Config File

Alternatively, create ~/.config/iqtoolkit-analyzer/config.json:

{
  "openai_api_key": "sk-...",
  "anthropic_api_key": "sk-ant-...",
  "gemini_api_key": "AI...",
  "aws_region": "us-east-1"
}

info

Environment variables take precedence over the config file.

Usage

# Use OpenAI for enhanced recommendations
iqtoolkit-analyzer analyze \
  --dsn "postgres://user:pass@localhost:5432/mydb" \
  --log-file /var/log/postgresql/postgresql.log \
  --ai-provider openai

# Use a specific model
iqtoolkit-analyzer analyze \
  --dsn "postgres://user:pass@localhost:5432/mydb" \
  --log-file /var/log/postgresql/postgresql.log \
  --ai-provider anthropic --ai-model claude-sonnet-4-20250514

# Set model via environment variable
export IQTOOLKIT_AI_MODEL="gpt-4o"
iqtoolkit-analyzer analyze \
  --log-file /var/log/postgresql/postgresql.log \
  --ai-provider openai

Sample AI Output

=== AI-Enhanced Recommendations ===
Based on your metrics, here are prioritized tuning suggestions:
1. Increase shared_buffers from 128MB to at least 1GB (25% of RAM)...
2. Set effective_cache_size to 75% of total RAM...
3. Consider enabling pg_stat_statements for query-level insights...