asc-localize-metadata
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interpolates untrusted metadata retrieved from App Store Connect into a translation prompt, which could allow malicious instructions within that metadata to influence the agent's behavior.
- Ingestion points: Metadata is ingested from external App Store Connect records via
asc localizations download(referenced in SKILL.md). - Boundary markers: The LLM translation prompt template uses triple quotes (
""") to delimit multi-line fields like descriptions and release notes. - Capability inventory: The skill performs shell command execution, file system reads/writes of
.stringsfiles, and network interactions through theascCLI. - Sanitization: No explicit sanitization or filtering of the ingested metadata is performed before it is interpolated into the prompt.
- [COMMAND_EXECUTION]: The skill frequently executes the
ascCLI tool to interact with App Store Connect APIs, including commands for listing resources, downloading files, and uploading modified metadata.
Audit Metadata