lobster-dev

What This Skill Does

The lobster-dev skill is a specialized development environment designed for engineers, researchers, and bioinformatics experts working with the Lobster AI multi-agent platform. It provides comprehensive tooling, documentation, and architectural patterns to help you build, extend, and debug complex bioinformatics agents. Whether you are modifying existing LangGraph-based orchestration, implementing new analysis services, or contributing to the core open-source infrastructure, this skill serves as your primary interface for maintaining high-quality code standards within the Lobster ecosystem.

Installation

To integrate the development environment into your workflow, run the following command in your terminal: clawhub install openclaw/skills/skills/cewinharhar/lobster-bio-dev Ensure you have the necessary development dependencies installed via make dev-install to enable full IDE support and testing capabilities within the repository structure.

Use Cases

• Agent Expansion: Developing new domain-specific agents such as genomics or proteomics experts and integrating them into the existing supervisor graph.
• Service Implementation: Building standardized analysis services that adhere to the mandatory 3-tuple return pattern (AnnData, Dict, AnalysisStep).
• Debugging & Maintenance: Identifying bottlenecks in agent workflows using the built-in provenance and logging tools provided by the core SDK.
• Architecture Exploration: Analyzing how tasks are routed through the LangGraph supervisor and how the ComponentRegistry handles agent discovery.

Example Prompts

1. "How do I register a new proteomics agent so the supervisor can discover it via the ComponentRegistry without hardcoding?"
2. "I need to implement a new transcriptomics analysis service. What is the standard pattern for returning the 3-tuple result?"
3. "Explain the current LangGraph architecture and how I should write unit tests for a custom tool in the lobster-genomics package."

Tips & Limitations

• Strict Namespace: Always follow the PEP 420 namespace structure; avoid creating __init__.py files in the root of the packages directory to prevent import conflicts.
• Efficiency: Keep your AGENT_CONFIG at the top of your modules to ensure agent discovery times stay under 50ms.
• Traceability: Always pass the ir (Internal Representation) object to the log_tool_usage function to maintain full audit logs of agent decision-making processes.
• Limitations: This skill assumes familiarity with Python, LangGraph, and bioinformatics data structures (specifically AnnData). It is a development-oriented tool and is not intended for end-user chat interactions without the accompanying local codebase clone.