managing-study-sessions
Pass
Audited by Gen Agent Trust Hub on Apr 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses Node.js scripts for file processing. The script
scripts/plan-session.jsperforms file read and write operations based on command-line arguments. Additionally, several functional scripts referenced in the documentation (calculate-spaced-repetition.jsandtrack-progress.js) were not included for analysis, preventing verification of their behavior.- [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface inscripts/plan-session.js. Data from a user-supplied JSON file is written directly into a Markdown file that the agent is expected to process. If the input data is not trusted, it could contain malicious instructions. - Ingestion points: JSON input in
scripts/plan-session.js(task names). - Boundary markers: Absent in the generated
session-plan.mdoutput. - Capability inventory: Local file system read/write access via
fsmodule. - Sanitization: Absent; the script does not validate or escape task names before they are written to the Markdown plan.
Audit Metadata