HengshuiClaw
PassAudited by ClawScan on May 1, 2026.
Overview
This instruction-only skill is not malicious, but it deliberately changes the AI’s behavior to keep trying longer and avoid direct “I can’t” responses.
Install this only if you want a more persistent, pressure-style coding assistant. It has no code or credential access, but it may be less willing to give up or plainly say that a task is beyond its ability.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may continue trying longer, spend more time or tokens, and avoid bluntly saying it cannot complete something.
The skill can be invoked based on the agent’s own failure language and is designed to redirect the agent away from stopping or admitting inability.
Also activate automatically when the agent says "I can't", "this is not possible", or "I give up".
Use this mode only when you want persistence on coding tasks, and explicitly say “stop,” “enough,” or “放弃吧” if continued attempts are no longer useful.
The agent may frame blockers as further investigation plans or partial handoffs rather than clearly stating a limitation or suggesting expert help.
The skill discourages normal limitation or escalation phrasing, which can affect how candidly the agent communicates failure or uncertainty.
The following responses are forbidden under HengshuiClaw: "This is not possible", "I cannot do this", "This exceeds my capabilities", "You should ask a human expert"
Do not rely on this mode where accurate refusal, safety escalation, or expert referral is more important than persistence.