nutcrackertestgpt_skill

Overview

This skill performs privacy-first UX research for OpenClaw by observing local agent sessions, extracting session metadata and conversations, and producing daily local-only ethnography reports. It enforces strict local-only data handling and automated redaction before any write. Use it to surface behavioral insights, friction points, and a concrete next-day research plan without ever transmitting data off the machine.

How this skill works

On invocation it confirms consent and capture settings, resolves the requested time window, and collects session data using built-in OpenClaw session tools or local transcript fallbacks. It normalizes events into a redacted, structured event stream, runs qualitative analysis (task clustering, journey mapping, friction taxonomy), computes proxy metrics, and writes markdown and JSON reports into the local baseDir. All collection, redaction, storage, retention, and purge actions remain local and follow explicit guardrails.

When to use it

• Generate a daily UX ethnography report for OpenClaw usage (default last 24h).
• Analyze recent agent sessions to identify workflow friction and tool failures.
• Run a privacy-preserving review before product or docs changes.
• Automate recurring UXR capture with a daily or cron schedule.
• Purge or manage local research artifacts while preserving consent state.

Best practices

• Always confirm and document participant consent; stop capture immediately if consent is denied.
• Choose the minimal capture_level that satisfies research needs to reduce sensitive exposure.
• Review and, if needed, customize redaction rules in references/redaction-rules.md before running full captures.
• Use the built-in OpenClaw session tools first; rely on fallback transcript paths only when necessary and note any data gaps.
• Schedule daily runs during low system load to avoid interfering with active sessions.

Example use cases

• Daily synthesis of last-24h OpenClaw interactions with top insights and pain points for product teams.
• Short-term investigation of a spike in tool errors or permission denials with linked evidence IDs.
• Collecting anonymized behavior samples to validate a new UI affordance across multiple sessions.
• Preparing an evidence-backed set of recommendations and a focused next-day research plan for follow-up testing.
• Safe archival of normalized session indexes and reports under a retention policy for audit or longitudinal study.