Security audit
qapten-sentient
Security checks for vulnerabilities and agentic risk
Overview
The skill's behavior (detecting emotion, reading/writing a single workspace file to persist tone state, and asking first-run consent) matches its stated purpose and there are no disproportionate requirements or installs.
This skill is coherent for its purpose: it reads and writes a single workspace file to remember a recent emotion-based tone and asks for permission on first use. Things to consider before installing: (1) the file stores an emotional calibration and could be considered sensitive — delete /home/node/.openclaw/workspace/memory/tone-state.md if you want to erase persisted state; (2) consent is only asked on first activation; if the file exists the skill will assume consent until you delete it or use the disable phrase; (3) although the skill's instructions say it won't auto-activate, the platform can invoke skills autonomously — review your agent's skill invocation settings if you need stricter control. If the skill started reading other files, requesting credentials, or posting to external endpoints, that would change this assessment to suspicious.
Static analysis
No suspicious patterns detected.