verification-before-completion
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill requires the agent to read and verify output from external commands and scripts, presenting a surface for indirect prompt injection if the output contains malicious instructions. * Ingestion points: Test outputs from shell commands (curl, drush eval) and local scripts in scripts/tests/. * Boundary markers: The skill does not define delimiters or provide instructions to ignore embedded instructions in test output. * Capability inventory: The agent is expected to use shell execution (curl, drush, ddev) and file system access (writing/reading scripts). * Sanitization: No sanitization or filtering of test output is specified.
- [COMMAND_EXECUTION]: The skill prescribes the use of command-line tools such as drush and curl to confirm task success. These are standard tools for the Drupal development stack identified in the skill metadata.
Audit Metadata