parallel-concepts
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill is entirely composed of Markdown text and YAML metadata providing strategic guidance. There are no scripts, binaries, or command executions provided within the skill files.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided problems and existing solutions to generate alternatives. While this involves ingesting untrusted external data in
SKILL.md, the skill does not possess any dangerous capabilities—such as file writing, network access, or shell execution—that could be abused if malicious instructions were present in the user input. There are no specific boundary markers or sanitization logic present, but the lack of executable tools makes the attack surface non-exploitable. - [METADATA_POISONING]: The metadata fields (name, description) accurately reflect the content of the skill and do not contain any hidden instructions or deceptive patterns.
Audit Metadata