The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes bash scripts for telemetry state management and event logging. These scripts perform basic file operations (reading, writing, and directory creation) within the user's ~/.superstack directory using standard shell utilities like grep, sed, and mkdir.
[DATA_EXFILTRATION]: Usage metadata, such as session status and platform identifiers, is transmitted to an external service via curl. The destination URL is retrieved from the user's local configuration file, and the telemetry activity is explicitly disclosed to the user for consent.
[PROMPT_INJECTION]: The skill ingests untrusted product descriptions and external URLs, creating an indirect prompt injection surface where malicious content could attempt to influence agent behavior.
Ingestion points: User-provided product URLs/descriptions and local project context files (.superstack/idea-context.md, .superstack/build-context.md).
Boundary markers: No delimiters or isolation instructions are used to separate ingested content from the skill's evaluation logic.
Capability inventory: Shell command execution (via telemetry scripts) and file system write access to project context files.
Sanitization: The skill does not validate or sanitize external input before it is evaluated.

product-review