earnings-task-setup

Warn

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill generates a Python command string using raw user-provided input from the $ARGUMENTS variable.
  • Evidence: python -c "from scripts.task_tracker import TaskTracker; TaskTracker().update_status('{TICKER}', '{QUARTER}', '{ACCESSION}', 'TASK_NAME', 'completed')"
  • Risk: This interpolation allows for command injection. A malicious user could provide a ticker or accession value containing escape characters (like single quotes and semicolons) to execute arbitrary Python or shell commands.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data without sufficient isolation or sanitization.
  • Ingestion points: User-supplied data enters the agent context via the $ARGUMENTS variable in SKILL.md.
  • Boundary markers: Absent. The skill does not use any delimiters or safety instructions to prevent the agent from obeying instructions embedded within the arguments.
  • Capability inventory: The skill utilizes Bash, Write, Read, and TaskUpdate tools.
  • Sanitization: Absent. No validation or escaping is performed on the variables {TICKER}, {QUARTER}, or {ACCESSION} before they are used in commands or file operations.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: User-controlled variables are used to determine local file storage paths.
  • Evidence: earnings-analysis/test-outputs/earnings-task-setup-{TICKER}-{QUARTER}.txt
  • Risk: This presents a path traversal risk if the user provides input containing sequences like ../, which could lead to writing files to unintended locations on the file system.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 21, 2026, 05:44 PM
Security Audit — agent-trust-hub — earnings-task-setup