learning-vault
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands (
mkdir -p,cp) to create directories and install plugin files into a user-specified vault location (defaulting to~/Brains/). This is used to set up the Obsidian environment as described in the skill's primary purpose. - [EXTERNAL_DOWNLOADS]: The skill distributes and installs a local copy of the 'Dataview' Obsidian plugin (authored by blacksmithgu). While the files are bundled with the skill rather than downloaded at runtime, this pattern of installing third-party executable code (
main.js) into a user's productivity application is noted. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external curricula via URLs provided by the user to auto-detect topics and domains. This creates an ingestion surface for untrusted data.
- Ingestion points: SKILL.md instructions for structure setup via URL curriculum.
- Boundary markers: None explicitly defined for the external content processing.
- Capability inventory: The skill has file-write capabilities (
mkdir,cp) and can execute shell commands to build the vault structure. - Sanitization: None described; the skill relies on the LLM to parse and structure the external curriculum into note files.
Audit Metadata