prep-plan
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local CLI tool (
questions.py) using user-provided input from thetarget_companiesvariable. The interpolation of this untrusted input directly into a shell command string (questions.py get "<Company>") creates a potential command injection vulnerability if the company name contains shell metacharacters like semicolons or pipes. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted historical data to generate future study plans, which creates an attack surface for instructions embedded in past session logs.
- Ingestion points: The skill reads historical user data and session logs from
$CRACKED_DIR/journal.md. - Boundary markers: Absent; there are no instructions to use delimiters or to ignore embedded instructions within the journal data.
- Capability inventory: The skill has the ability to execute shell commands (
date,python3,questions.py) and write files ($CRACKED_DIR/plan.md). - Sanitization: Absent; while the skill normalizes tags for frequency counting, it does not sanitize the broader content that influences the strategy and narrative of the generated plan.
Audit Metadata