Custom Engine Configuration
GoSiteMe's engine architecture is modular—each engine (ELEPHANT, PLAYBOOK, CLOCKWORK, RAG, and others) can be configured independently to match your project's requirements. This guide explains how to customize engine settings, chain engines together, and tune performance.
Accessing Engine Settings
- Go to Dashboard → Engines.
- Click on the engine you want to configure.
- The settings panel shows all configurable parameters for that engine.
ELEPHANT Engine Configuration
ELEPHANT handles persistent memory and context management. Key settings:
- Memory retention period: How long context is preserved (session, 24 hours, 7 days, permanent).
- Context window size: Maximum token count for context retrieval (4K, 8K, 16K, 32K).
- Priority weighting: Assign higher priority to recent interactions vs. older knowledge.
- Compression: Enable context compression to fit more information within the token limit.
PLAYBOOK Engine Configuration
PLAYBOOK manages workflow templates and automated sequences:
- Execution mode: Sequential (one step at a time) or parallel (independent steps run simultaneously).
- Retry policy: Number of retry attempts and backoff interval for failed steps.
- Timeout: Maximum execution time per step and per workflow.
- Logging level: Debug, info, warn, or error.
CLOCKWORK Engine Configuration
CLOCKWORK handles scheduled tasks and time-based triggers:
- Timezone: Set the default timezone for all scheduled jobs.
- Concurrency limit: Maximum number of jobs running simultaneously.
- Missed job policy: Run immediately, skip, or queue when a scheduled job is missed.
- Notification channel: Where to send alerts for completed or failed jobs.
Engine Chaining
Engines can be chained so the output of one feeds into the next. Configure chains under Engines → Chains:
- Click Create Chain.
- Add engines in the desired sequence (e.g., RAG → ELEPHANT → PLAYBOOK).
- Define data mappings between engine outputs and inputs.
- Set a trigger (manual, scheduled, or event-based).
- Save and activate the chain.
Performance Tuning
- Monitor engine resource usage under Engines → Monitoring.
- Increase the allocated compute for engines processing heavy workloads.
- Use caching settings to avoid reprocessing identical requests.
- Review engine logs for bottlenecks and optimize slow steps in your chains.