Multi-viewpoint Debates

Spawn isolated sub-agents representing distinct worldviews (Elon, Capitalist, Monkey) to debate decisions from multiple angles. Expose blind spots by forcing genuine disagreement on important questions. Use when facing decisions where you need to challenge your assumptions, stress-test ideas, or see a problem through fundamentally different lenses. Automatically captures debate outputs to an archive for future reference and pattern analysis.

This skill appears coherent and safe from a capability/requirement standpoint, but review these before installing: 1) It runs clawdbot sessions_spawn commands — confirm you want the agent to spawn sub-agents on your system. 2) It expects you to collect and save session transcripts (examples point to ~/.clawdbot/agents/...), so archives may contain sensitive project details—store them securely and check their contents before sharing. 3) The skill uses public-figure persona labels (e.g., "Elon") — be aware of any platform or legal policy on imitating real people. If you want to proceed, verify the filesystem paths in the docs match your environment and that you are comfortable allowing Clawdbot to create and store debate markdown files locally.

Type: OpenClaw Skill Name: multi-viewpoint-debates Version: 1.0.0 The skill is designed to spawn isolated sub-agents for multi-viewpoint debates. The `scripts/run-debate.sh` script reads local persona definitions and constructs `clawdbot sessions_spawn --task` commands, which are then printed for the user to execute. While user input (the 'TOPIC') is embedded into the sub-agent's prompt, this is a standard prompt injection vector inherent to AI systems and is not indicative of malicious intent from the skill itself, especially since the persona framework is prepended. There is no evidence of data exfiltration, malicious execution, persistence, or obfuscation. All file operations are local and within the skill's expected scope.