lobster-bio-dev
por cewinharhar
Coding Agents & IDEsResumen
Develop, extend, and contribute to Lobster AI — the multi-agent bioinformatics engine.
Descripción
Qué hace esta skill
A development guide for Lobster AI, a multi-agent bioinformatics platform built on LangGraph. Covers adding agents and services, navigating the orchestration graph, and contributing to the open-source codebase. Key conventions include entry-point-based agent discovery and a mandatory 3-tuple service return pattern.
The skill captures non-obvious codebase conventions like AGENT_CONFIG placement, entry-point registration, and the mandatory ir parameter that are not apparent from reading the source alone.
Cuándo usarla
- Adding a new domain expert agent for a bioinformatics modality
- Implementing analysis services with the required 3-tuple return
- Tracing query flow from CLI through LangGraph to agent output
- Writing unit tests for new agents before submitting a PR
- Debugging agent discovery failures from missing entry point registration
Flujo de trabajo de ejemplo
User asks: 'Add a new metabolomics expert agent to Lobster'
- 1
Creates packages/lobster-metabolomics/ following PEP 420 namespace structure
- 2
Defines AGENT_CONFIG at the top of the agent file, before any heavy imports
- 3
Registers the agent via entry point in pyproject.toml under lobster.agents
- 4
Implements tools wrapping services with 3-tuple return and log_tool_usage(ir=ir)
- 5
Runs pytest tests/unit/agents/ to verify registration and behavior
A new metabolomics_expert agent discoverable by ComponentRegistry and routable via the supervisor
Requisitos
- Anthropic API key (ANTHROPIC_API_KEY) for running agents during development and testing
Instalación
clawhub install cewinharhar/lobster-bio-devEjecuta el comando de la CLI o pega el prompt en Claude Code, Cursor, Codex o el agente de programación que uses.