vault-maintenance
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts, including check_vault.py and generate_knowledge_index.py, to automate validation and indexing tasks. These are executed via python3 as part of the maintenance workflow.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes Markdown files from the notes/, references/, and .agents/ directories, which creates a potential surface for indirect prompt injection if notes contain malicious instructions.
- Ingestion points: Markdown files across the vault's core directories are parsed by the maintenance scripts.
- Boundary markers: The scripts distinguish between frontmatter and note bodies but do not implement specialized delimiters to prevent the agent from interpreting note content as active instructions.
- Capability inventory: The skill is capable of file system modifications (move, rename, write) and local script execution within the agent's operating environment.
- Sanitization: Note bodies are read and processed for metadata and consistency checks without additional sanitization or instruction filtering logic.
Audit Metadata