spec-execute

Pass

Audited by Gen Agent Trust Hub on Jul 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands extracted from {FEATURE_DIR}/implementation/plan.md and task.md. This is the core functionality but allows for arbitrary command execution if those files contain malicious content.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by following instructions embedded in project files which are treated as the Single Source of Truth (SSOT).
  • Ingestion points: {FEATURE_DIR}/implementation/plan.md and {TASK_PACK_DIR}/task.md.
  • Boundary markers: None. The skill does not utilize delimiters or warnings to differentiate between task data and executable instructions.
  • Capability inventory: The agent can execute shell commands, perform Git operations, and modify the local file system.
  • Sanitization: No evidence of sanitization or validation of the commands or instructions parsed from the external plan files is present.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 28, 2026, 08:08 AM
Security Audit — agent-trust-hub — spec-execute