obsidian-wiki-init
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a Python script to perform environment discovery and vault initialization tasks. Evidence includes the use of python commands to invoke a runtime script for root discovery and state management.
- [REMOTE_CODE_EXECUTION]: The skill invokes an external script located at a computed path relative to the skills root. This logic relies on the presence of llm_wiki.py in the obsidian-wiki-runtime directory.
- [EXTERNAL_DOWNLOADS]: The skill instructs users to install missing components from the huajiexiewenfeng/obsidian-llm-wiki repository using the npx skills add command, which fetches and executes external code.
- [PROMPT_INJECTION]: The skill processes untrusted vault content (filenames and metadata) to generate summaries and roadmaps, creating an indirect prompt injection surface. The ingestion occurs during vault scanning, with boundary markers established via safety instructions that limit deep scanning and secret copying. The skill maintains the capability to execute commands via its runtime script, and no explicit sanitization of vault data is mentioned.
Audit Metadata