kernel-extract-posts
Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or hardcoded credentials detected in the instructions.
- [SAFE]: The use of the Read and Write tools is consistent with the skill's stated purpose of reading an essay and saving social media content.
- [PROMPT_INJECTION]: Potential surface for indirect prompt injection identified: 1. Ingestion points: Reads untrusted content from a user-specified essay file via the Read tool. 2. Boundary markers: No explicit delimiters are used to wrap the input essay content or instruct the model to ignore embedded directives. 3. Capability inventory: The skill has Write access to the local filesystem. 4. Sanitization: No validation or filtering is applied to the input text. Evaluation: The risk is categorized as safe because the instructions force the output into a narrow JSON schema, significantly limiting the impact of malicious content in the processed essay.
- [NO_CODE]: This skill contains instructions only and does not include any scripts or binaries.
Audit Metadata