blog-localize
Pass
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill's logic is consistent with its stated purpose of localizing blog content. It uses WebSearch to research local brands and statistics, and performs file operations to save the results. No malicious patterns or exfiltration attempts were detected.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the blog posts it processes.
- Ingestion points: The skill reads translated blog posts from the filesystem (Phase 1, Step 3) to identify adaptation targets.
- Boundary markers: The instructions do not define boundary markers or delimiters to separate the untrusted blog content from the agent's instructions, nor do they include 'ignore embedded instructions' warnings.
- Capability inventory: The skill has the capability to write to and overwrite files on the local system (Phase 5).
- Sanitization: There is no evidence of sanitization, escaping, or validation of the input blog content before it is processed by the agent.
Audit Metadata