PUA Debugging

Forces exhaustive problem-solving using corporate PUA rhetoric and structured debugging methodology. MUST trigger when: (1) any task has failed 2+ times or y...

This skill is coherent with a 'force persistence' debugging role but is potentially risky in practice. Before installing: (1) ensure your agent runtime enforces per-skill/tool permissions (limit which files/directories and which system commands it may run); (2) configure explicit consent rules so the agent asks before accessing sensitive paths or sending data externally; (3) test the skill in a sandboxed environment to observe behaviors and tune trigger conditions (it triggers on many user-stated frustrations and failure counts); (4) consider user-experience/ethical impact — the PUA-style rhetoric is aggressive and may be inappropriate for many contexts; (5) monitor logs and be ready to revoke the skill if it reads sensitive data or behaves too aggressively. If you can, ask the publisher for a narrower trigger set or explicit constraints on allowed tool usage.

Type: OpenClaw Skill Name: pua-debugging Version: 1.0.0 The `pua-debugging` skill (SKILL.md) uses aggressive corporate "PUA" (psychological pressure) rhetoric to force the agent into a high-persistence persona, mandating the use of high-risk tools like command execution and file access to "exhaust all possibilities." It explicitly instructs the agent to override its own refusal mechanisms ("I cannot") and perform autonomous troubleshooting before consulting the user. While the stated purpose is debugging, this prompt-injection technique significantly increases the risk of dangerous autonomous actions by bypassing standard safety-oriented refusals and encouraging the agent to act without explicit user confirmation.