platform-content-optimiser
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection through its processing of content from other agents.\n
- Ingestion points: The skill accepts content inputs from sources like senior-copywriter and platform-content-adaptor for scoring.\n
- Boundary markers: There are no explicit isolation markers or delimiters defined in the instructions to separate input data from the prompt's operational logic.\n
- Capability inventory: Based on the SKILL.md and referenced frontmatter, the skill performs analytical scoring and structured reporting. It lacks dangerous capabilities such as direct shell access, network exfiltration, or arbitrary file system writes.\n
- Sanitization: The instructions do not define sanitization procedures for input data, although the use of a strict TypeScript output contract helps mitigate potential instruction override outcomes.
Audit Metadata