oc-social-agents

Summary

This workspace provides agent configuration files and core behavior documents for an OpenClaw-powered social media management system. It defines roles, workflows, personalities, and delegation rules for a multi-agent setup designed to automate content planning, research, drafting, and publishing on social platforms.

Included Assets

• Agent configuration & behavior files:
  • AGENTS.md: Workspace overview, session, and memory guidelines
  • SOUL.md: Main agent (Muddy) identity and routing logic
  • TOOLS.md: Local environment notes
• Agent directories and documentation:
  • agents/README.md: Multi-agent architecture overview
  • agents/prime/README.md, agents/pulse/README.md, agents/scout/README.md, agents/quill/README.md, agents/twitter/README.md: Role definitions and workflow for each agent
  • agents/prime/delegation-rules.md: Model assignments and spawn templates
• Agent "SOUL.md" files:
  • Personality, values, and decision style for Prime, Pulse, Scout, Quill, Twitter agents
• Node dependency:
  • node_modules/dotenv/ and package.json: For managing environment variables

How to Use

1. Review agent role files:
   • Start with agents/README.md for system architecture.
   • Read each agent’s README.md and SOUL.md for personality, tasks, and workflow specifics.
2. Customize workspace notes:
   • Add environment-specific details to TOOLS.md.
   • Update AGENTS.md as needed to track sessions and memory.
3. Follow assignment and delegation rules:
   • Refer to agents/prime/delegation-rules.md for proper model assignment and agent spawning syntax.
4. Set up environment variables (optional):
   • Install dependencies via npm install.
   • Use dotenv (.env file) for sensitive configs if needed.
5. Operate the agents as described:
   • Maintain the workflow pipeline: Pulse triggers → Prime plans → Scout researches → Quill drafts → Twitter posts.
   • Always route tasks through designated agents; avoid role confusion.

Notes

• Each agent is strictly scoped—refer to their docs for what they do and don’t do.
• Memory and session instructions in AGENTS.md help maintain continuity and context for agent operation.
• Keep environment-specific items (devices, aliases) in TOOLS.md.
• Agents do not share confidential context unless approved for main session access.
• Content publishing is handled stepwise—never skip the research or draft approval process.
• For any updates, check the relevant SOUL.md or delegation rules to avoid disrupting the workflow.