lobster-agent-forge

What This Skill Does

The lobster-agent-forge skill is the primary architectural toolkit within OpenClaw for deploying, managing, and orchestrating autonomous AI agents. Unlike standard chatbots, agents built with this forge are designed for persistence and independent operation. The skill provides the scaffolding, organizational structure, and cron-based orchestration required to build agents that possess a 'soul' (identity), a 'heart' (rhythm of execution), and a 'memory' (long-term knowledge retention). By leveraging the agent generator script, you can initialize standardized directory structures that ensure your agents remain coherent across multiple sessions, allowing them to evolve, refine their decision-making processes, and execute complex workflows without constant human intervention.

Installation

To integrate the lobster-agent-forge into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/crispyangles/lobster-agent-forge

Once installed, you can trigger the creation of a new agent immediately using the provided generator script. Ensure you have the necessary environment permissions to write files and execute scripts, as the forge will initialize folders and core documentation files (SOUL.md, AGENTS.md, etc.) in your designated directory.

Use Cases

• Autonomous Market Research: Build an agent that runs nightly, scans market trends, updates your MEMORY.md file, and provides a summary report in the morning.
• Multi-Agent Orchestration: Create a 'Manager' agent that delegates specific tasks to specialized 'Worker' agents, coordinating their HEARTBEAT.md activities for team-based operations.
• Continuous Task Management: Set up a dedicated bot to monitor project status, check for incomplete items in your project files, and proactively remind you of upcoming deadlines.
• Self-Evolving Persona: Develop an agent that reflects your specific communication style and business priorities, using the SOUL.md framework to ensure it adheres to your strict operational constraints and guardrails.

Example Prompts

1. "Create an agent named 'Strategist' in the ./agents/strategist directory to manage my daily stock market updates."
2. "I need to configure a new heartbeat for my social-media-bot that checks for pending posts every 30 minutes between 9am and 5pm."
3. "Set up an agent persona for a project management bot that uses a sarcastic tone and prioritizes high-revenue tasks."

Tips & Limitations

• File-First Persistence: Always remember that agents clear their active context on restart. If information is critical, it must be saved to MEMORY.md or a specific log file.
• Cron Precision: Use the --tz flag when setting up crons to ensure your autonomous tasks occur in your local time zone.
• Guardrail Definition: Be highly specific in your SOUL.md. Ambiguous goals lead to drift; define clear 'Do' and 'Never do' lists to maintain control over autonomous behavior.
• Resource Usage: Because agents execute autonomously, monitor your cron logs to ensure they aren't performing redundant operations that could lead to unnecessary file bloat or resource consumption.