Curiosity Engine
PassAudited by ClawScan on May 10, 2026.
Overview
This skill is a transparent reasoning aid, but it may encourage extra web, file, command, and optional memory use during deeper analysis.
This appears safe to install as a reasoning enhancement. Be aware that it can make the agent more exploratory: it may perform extra searches, read files, or run local commands when those tools are available. Keep approval controls on for file and command tools, and avoid enabling memory for sensitive topics unless you are comfortable retaining them.
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.
For deeper questions, the agent may spend extra time and may use available web, file, or command tools to verify information.
The skill explicitly authorizes external research tools and local read/exec tools during analysis. This is aligned with its research-enhancement purpose and is bounded, but users should understand it may trigger extra tool use.
- Use web_search, web_fetch, read, exec as appropriate - Prioritize: highest information gain first, max 3 tool explorations per loop
Keep normal tool approval prompts enabled for file reads and exec commands, and use the provided `/curious off` or quick-answer instructions when you do not want extra exploration.
Topics or questions from prior sessions could be saved and reused later if memory is enabled.
The skill may retain future research questions across sessions, but it is disclosed and conditioned on user opt-in.
Store persistent open threads in `memory/curiosity-threads.md` if the user opts into memory.
Only enable memory if you are comfortable retaining these threads, and review or delete `memory/curiosity-threads.md` if it contains sensitive topics.