kimi-slides
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted content from user-uploaded files and provided URLs to generate presentation content.
- Ingestion points: The
SKILL.mdfile instructs the agent to "Read all files uploaded by the user, the provided URLs" as the first step in the production workflow. - Boundary markers: The instructions do not specify the use of boundary markers (e.g., XML tags or clear delimiters) or provide warnings to ignore embedded instructions within user data, which could allow maliciously crafted input to influence the agent's behavior.
- Capability inventory: The skill has access to a companion CLI tool (
kimi-slides) and allows the agent to executebashorpythoncommands for image manipulation (e.g., cropping and screenshotting), creating a path for downstream impact if injected content influences these operations. - Sanitization: No specific sanitization or filtering logic for the ingested content is described in the provided files.
Audit Metadata