wordpress-workspace-transcription-smoke-test
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local POSIX shell script located at 'Tools/wpcom-transcribe.sh'. While this script is internal to the skill's infrastructure, its contents were not provided for direct verification.
- [PROMPT_INJECTION]: The skill processes untrusted input in the form of transcription text and application context, which creates an attack surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context via the '--selected-text' and '--app-context' arguments passed to the transcription tool as described in 'SKILL.md'.
- Boundary markers: The instructions do not define any delimiters or negative constraints (e.g., 'ignore embedded instructions') to prevent the agent from obeying instructions contained within the transcribed text.
- Capability inventory: The skill has the capability to execute shell scripts and perform network operations via 'curl' to WordPress.com API endpoints.
- Sanitization: There is no mention of validation, escaping, or filtering for the external text or JSON data before it is processed by the agent.
Audit Metadata