baoyu-xhs-images
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text to create infographic outlines and image prompts, creating a surface for indirect injection attacks.
- Ingestion points: User-provided content is saved to
source-{slug}.{ext}as described inSKILL.md. - Boundary markers: The prompt assembly process in
references/workflows/prompt-assembly.mddoes not use explicit delimiters or isolation markers when interpolating user content into final prompts. - Capability inventory: The skill has the capability to write files, invoke image generation tools, and execute shell commands via local wrappers (referenced in
SKILL.mdandreferences/codex-imagegen.md). - Sanitization: No sanitization, escaping, or validation of user-provided content is described before it is used in prompt construction.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to interface with image generation backends.
- Evidence:
references/codex-imagegen.mdprovides specific instructions for executingbunornpxto run thebaoyu-image-genorcodex-imagegenwrapper scripts. - [DYNAMIC_EXECUTION]: The skill dynamically resolves and executes local scripts at runtime based on environment variables and directory discovery.
- Evidence: In
references/codex-imagegen.md, the agent is instructed to discover thebaoyu-codex-imagegenwrapper by checking environment variables (BAOYU_CODEX_IMAGEGEN_BIN) or searching the plugin root directory, then executing the resulting path.
Audit Metadata