strategic-decision

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to manage a local directory (~/.strategic-decisions) for storing and retrieving strategic decision records. This includes creating the directory and listing existing files to find related context.
  • [PROMPT_INJECTION]: The skill processes user-supplied decision topics and incorporates them into shell commands for filename generation, creating an indirect prompt injection surface. However, the risk is addressed through mitigation logic.
  • Ingestion points: Strategic decision topics and context provided by the user in SKILL.md.
  • Boundary markers: Absent; user-provided data is interpolated into subsequent processing steps.
  • Capability inventory: Filesystem interaction and command execution (mkdir, ls, sed, tr, date) via agent tools.
  • Sanitization: Present; the skill uses tr and sed to sanitize user-provided topics, removing non-alphanumeric characters before using them in shell commands for filename creation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 05:32 PM
Security Audit — agent-trust-hub — strategic-decision