trace_skill

This skill analyzes real user sessions to uncover why actions happened, linking persona insights with narrative UX recommendations.

Shell

8

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill simota/agent-skills --skill trace

SKILL.md5.8 KB

Overview

This skill analyzes real user session replays to extract persona-based behavior patterns and tell the UX story behind metrics. It acts as a behavioral archaeologist: segmenting sessions, detecting frustration signals, reconstructing journeys, and producing evidence-backed recommendations. Outputs are designed to validate personas and guide simulations and visualizations.

How this skill works

Trace ingests anonymized session logs and replay events, then segments sessions by persona or behavior cohorts. It runs signal detection (rage clicks, back loops, scroll thrash, dead clicks, form abandonment) and scores sessions to surface high-friction clusters. The agent reconstructs user journeys as narrative reports, quantifies patterns, and prepares structured handoffs for researchers, simulators, and visualization tools.

When to use it

Investigating sudden metric drops (conversion, engagement) flagged by analytics
Validating or updating research personas against real behavior
Finding root causes of mobile or platform-specific navigation problems
Preparing evidence and scenarios for simulation or A/B test design
Producing narrative reports for stakeholders that need context beyond charts

Best practices

Always confirm access and privacy constraints before analyzing session replays
Provide persona definitions or ask for researcher input when segmenting
Prioritize high-severity signals (rage clicks, back loops) but verify patterns with sample sessions
Cite anonymized evidence and quantify sample sizes to avoid overclaiming
Handoff findings with clear next actions and suggested agents (Researcher, Echo, Canvas)

Example use cases

Pulse flags a 15% conversion drop → Trace examines recent sessions to reveal a checkout step causing rage clicks and abandonment
Researcher defines a novice persona → Trace validates the persona by finding consistent hesitation and help-seeking in that cohort
Echo predicts a friction point in onboarding → Trace checks replays to confirm whether real users show matching behavior
Product team wants a story-driven report → Trace delivers a narrative with annotated replay excerpts and prioritized recommendations
Design needs journey diagrams → Trace exports segmented flow data to Canvas for visualization

FAQ

Anonymized session logs or replay streams, event timestamps, and optional persona filters. Confirm privacy requirements before access.

How does Trace measure frustration?

Using a signal taxonomy: rage clicks, back loops, scroll thrash, dead clicks, form abandonment and a composite score that prioritizes high-severity patterns.