siyu-save
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a shell command via python3 to generate a standardized ISO 8601 timestamp. This is used exclusively for internal metadata and file naming, and the command string is static, preventing shell injection.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted dialogue content to generate summaries, which presents a surface for potential indirect prompt injection.
- Ingestion points: The skill ingests conclusions and summaries from the current dialogue history.
- Boundary markers: Extracted data is placed into a structured Markdown schema with predefined frontmatter fields and section headers.
- Capability inventory: Includes file system write access to a specific local directory (~/.siyu/clients/) and local shell execution for timestamping.
- Sanitization: Implements a rigorous sensitive information scanning process for PII and credentials (API keys, tokens). It requires a manual review step where the user must explicitly choose between redacted or original saving before any file is created on disk.
Audit Metadata