Back to skill
Skillv1.0.4
ClawScan security
Elicitation - how to talk with humans and ask them questions? · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignFeb 11, 2026, 8:58 AM
- Verdict
- benign
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill is an instruction-only elicitation guide whose requested resources and runtime instructions match its stated purpose, but it raises important privacy and ethical considerations for use.
- Guidance
- This skill is internally coherent but deals with sensitive human-subject material. Before installing or using it, consider: obtain clear, informed consent from people you question; avoid using the skill for diagnosis, clinical treatment, or with vulnerable populations (minors, people in crisis) unless you are qualified and have proper safeguards; set explicit limits on topic depth and number of follow-ups so the 'iterate until 10/10' instruction does not lead to persistent probing; log and store any collected personal data securely and minimize retention; be transparent with users about analytic inferences and uncertainty (language-based inferences are probabilistic); and ensure compliance with any applicable privacy laws or organizational policies. If you plan to use it in products or research, add explicit consent workflows, escalation paths for disclosure of harm, and review by an ethics or legal team.
Review Dimensions
- Purpose & Capability
- okThe name and description (psychological elicitation / narrative identity / MI) align with the included SKILL.md and supporting markdown files. There are no unexpected environment variables, binaries, or installs required that would be inconsistent with a conversational elicitation tool.
- Instruction Scope
- noteThe SKILL.md and companion files are instruction-only and direct the agent to use OARS, self-defining-memory prompts, schema detection, LIWC-style language inference, and structured life-review sequences. They do not instruct the agent to read system files, environment variables, or external endpoints. One behavioral instruction — 'iterate until you reach 10/10' — gives broad operational latitude and could encourage repeated probing or prolonged interaction; this is within the skill's purpose but can lead to scope creep in practice and needs human oversight and explicit consent from interlocutors.
- Install Mechanism
- okNo install spec and no code files to execute. This is the lowest-risk pattern for disk writes or remote code fetch. The regex scanner had no files to analyze beyond static markdown guidance.
- Credentials
- okThe skill declares no required environment variables, credentials, or config paths. There are no requests for unrelated secrets or host resources; requested access is proportionate (none).
- Persistence & Privilege
- okSkill flags are default (always:false, agent invocation allowed). It does not request permanent presence or permissions to modify other skills or system settings.