The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection because it processes session data without sufficient sanitization before rendering.
Ingestion points: Reads conversation history from JSONL logs in ~/.claude/projects/ within both token_calendar.py and token_stats.py.
Boundary markers: Absent; log content is directly interpolated into HTML templates via JSON replacement.
Capability inventory: Generates HTML reports in temporary directories and opens them in a browser via subprocess.run().
Sanitization: Absent; the frontend JavaScript uses innerHTML to display model names and project identifiers, creating a Cross-Site Scripting (XSS) vector if session logs contain malicious payloads.
[COMMAND_EXECUTION]: The scripts execute system utilities (open, xdg-open) using subprocess.run() to display the generated visualizations to the user.
[EXTERNAL_DOWNLOADS]: Fetches visualization libraries (Chart.js) from the jsDelivr CDN and font styles from Google Fonts to enhance the report presentation.

token-stats