What problem does it solve? Contributing to a large (~144K lines) Python agent orchestration platform without knowing its architecture leads to inconsistent, incorrect changes. This Skill provides the architectural context, patterns, and development rules needed to modify PawFlow correctly. ## Core Features & Use Cases - Architecture Orientation: Explains the two major subsystems (NiFi-inspired pipeline engine and multi-agent LLM system) and the full directory layout of core/, tasks/, services/, and engine/. - Pattern Enforcement: Documents key conventions such as thread-safe singleton stores, JSONL append-only conversation persistence, relay-backed tool handlers, and the ${scope.key} expression language. - Development Rules: Lists concrete contribution rules covering tests, docs, parameter validation, async actions, and force-stop semantics. - Use Case: When asked to add a new tool handler to PawFlow, the agent knows to place it in core/handlers/, register it via ToolRegistry, route execution through the relay layer, and add focused tests. ## Quick Start Load the pawflow-developer skill and help me add a new tool handler to the PawFlow codebase following its conventions.
