planning-coaching

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on a local command runner ('just') located in the user's home directory (/home/imalison/org/justfile) to interact with a task management system. It executes various subcommands to create, update, and complete tasks.
  • [DATA_EXFILTRATION]: The skill accesses and modifies personal data in the user's home directory (/home/imalison/org/planning/), including context files about the user's life and daily journals. While this involves sensitive personal information, it is the primary purpose of the skill.
  • [PROMPT_INJECTION]: An indirect prompt injection surface exists because the skill is instructed to read content from user-controlled files (context.org and daily journals) at the start of every session without sanitization.
  • Ingestion points: /home/imalison/org/planning/context.org, /home/imalison/org/planning/dailies/YYYY-MM-DD.org
  • Boundary markers: Absent; the skill is directed to read these files directly into context.
  • Capability inventory: The skill can execute local commands (just), write/append to files, and modify task states.
  • Sanitization: Absent; no validation or filtering is applied to the content of the persistent files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 01:28 AM