Skip to content

README.md

Latest commit

 

History

History
45 lines (31 loc) · 1.28 KB

README.md

File metadata and controls

45 lines (31 loc) · 1.28 KB
Raw

Generic Plugin

Direct LLM invocation with optional chat history condensation.

Overview

Property Value
Plugin type generic
Event type Input
Queue virtual-contributor-engine-generic
Ports LLMPort

How It Works

The simplest plugin in the system. Routes user messages directly to the LLM with no retrieval step.

Input event
  → (if history present) Condense conversation into standalone question via LLM
  → Build message array from system prompt + condensed question
  → LLM invocation
  → Response (no sources)

When conversation history is present, prior exchanges are condensed into a standalone question before the final LLM call. This prevents context window bloat from long conversation threads.

Configuration

No plugin-specific configuration. Per-plugin LLM overrides are supported via GENERIC_LLM_* prefix.

The plugin also supports per-request engine selection via input.engine and external_config.api_key for dynamic LLM routing.

Key Files

File Purpose
plugin.py Plugin implementation — history condensation and LLM invocation
prompts.py Condensation prompt template

Testing

poetry run pytest tests/plugins/test_generic.py