skills/wintree86/plan-task-fix/fix/Gen Agent Trust Hub

fix

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes git rev-parse --show-toplevel to programmatically identify the project's root directory. This is a legitimate and safe operation used to determine the correct location for the backlog file.\n- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests user-provided text as task descriptions and stores them in backlog.md, which the agent later reads. This behavior is consistent with the skill's primary purpose and is classified as safe.\n
  • Ingestion points: User input through the /fix command arguments and existing markdown content in backlog.md.\n
  • Boundary markers: None present; input is directly interpolated into markdown list items.\n
  • Capability inventory: Reads and writes to the local backlog.md file and executes basic Git commands for path resolution.\n
  • Sanitization: No explicit sanitization or escaping of the description text was found before it is written to the file.\n- [SAFE]: No obfuscation, hardcoded credentials, or unauthorized network operations were detected. The skill's external reference to the author's repository is documented as a trusted vendor resource.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 11:30 PM