Glossary — quick reference

• .ww: The declarative configuration file format used by Woodwork Engine (e.g., main.ww). It declares components and how they connect.

• Component: A modular unit (LLM, agent, input, output, tool, etc.) defined in a .ww file and instantiated by the parser.

• Parser: The code that reads .ww files and instantiates configured components (see woodwork/parser/config_parser.py).

• Task Master: The orchestrator that manages workflows and component interactions (woodwork/core/task_master.py).

• Agent: A component that performs reasoning or decision-making, often backed by an LLM (e.g., agent openai { ... }).

• LLM: Large Language Model component (OpenAI, HuggingFace, etc.) used by agents to generate text.

• Knowledge Base (KB): A component that stores or indexes data for retrieval (e.g., Chroma, text files).

• Event: Typed messages exchanged between components (e.g., input.received, agent.thought). Events are JSON-serializable and typed in woodwork/types/events.py.

• Payload: Structured data carried by an event. Payload types are defined under woodwork/types/ and validated by the PayloadRegistry.

• Hook: A read-only listener that runs when events occur (used for logging, monitoring). Hooks do not modify payloads.

• Pipe: A transform that runs sequentially on an event and can modify its payload before it continues through the system.

• Workflow / Pipeline: A configured flow connecting components so events move from inputs → agents → outputs/tools.

• CLI Input: The built-in command-line input component used for quick local testing (input.cli).

• Example / examples/: Directory with runnable .ww examples. Copying and adapting these is the fastest way to learn.

• main.ww: Common filename used as the primary entrypoint configuration for a Woodwork project.

Where to learn more

• Quickstart: docs/quickstart.md — the shortest path from install → running an agent.
• Beginner overview: docs/beginner-overview.md — architecture summary and next steps.
• Tutorials: docs/tutorials/first-project.md — minimal walkthrough creating and running a tiny project.

If a term here is unclear or missing, please open a PR to expand this glossary — small clarifications help new contributors a lot.