solo-swarm

What This Skill Does

The solo-swarm skill transforms your OpenClaw agent into a project lead, orchestrating a team of three specialized research agents that work in parallel. Instead of relying on a single linear research process, solo-swarm delegates tasks to three distinct personas: a Market Researcher, a User Researcher, and a Technical Analyst. This multi-threaded approach allows for a holistic investigation of any idea, ensuring you cover commercial viability, user sentiment, and technical feasibility simultaneously. The skill culminates in a synthesized research.md file that provides a clear recommendation and a roadmap for your next steps, such as validation.

Installation

To integrate solo-swarm into your workflow, ensure your environment is prepared by setting the CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS environment variable to 1 in your .claude/settings.json file. Once configured, you can install the skill by running the following command in your terminal:

clawhub install openclaw/skills/skills/fortunto2/solo-swarm

Use Cases

This skill is perfect for entrepreneurs, product managers, and developers during the early "discovery phase" of a project. Use solo-swarm when you need to:

• Validate a new startup concept or feature idea.
• Pivot or explore an alternative market for an existing product.
• Rapidly assess the "buy, build, or abandon" potential of a technical idea.
• Gain a competitive advantage by analyzing market gaps and user frustrations without spending weeks on manual research.

Example Prompts

1. "Swarm research: I want to build a decentralized platform for freelance contract management. Check if this is viable."
2. "Can you perform a parallel research task on the viability of an AI-powered personal finance tracker for Gen Z?"
3. "Run a team research investigation on integrating local LLMs for offline document processing. I need a tech and user perspective."

Tips & Limitations

• Focus on Separation: The power of this skill lies in the differentiation of roles. If you notice findings are overlapping, encourage the agents to stick strictly to their assigned domain (Market, User, or Tech).
• Prerequisites: Always verify that your project directory has write permissions, as the skill generates a research.md file.
• When Not to Use: Avoid using this for simple questions or solo deep dives. If you only need one perspective, use the standard /research command to save tokens and time. For confirming if an idea is "good" based on specific metrics, use /validate instead of this exploration-focused tool.
• Configuration: If the agent team fails to initialize, double-check your environment variables; the team-based architecture requires specific runtime permissions to spawn multiple processes.