obsidian-wiki-ingest

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill dynamically determines the path for its core logic script (llm_wiki.py) and executes it through the Python interpreter to perform tasks like vault resolution and document ingestion.
  • [EXTERNAL_DOWNLOADS]: If the required runtime is not found locally, the skill provides the user with an installation command (npx skills add huajiexiewenfeng/obsidian-llm-wiki) to fetch the package from a remote repository. This resource is associated with the skill's author.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and summarizing external, untrusted documents such as PDF and Markdown files into the Obsidian vault.
  • Ingestion points: Files and external directories processed during the scan and ingest workflows (defined in SKILL.md and references/ingest-workflow.md).
  • Boundary markers: Not explicitly mentioned; generated source proxy nodes and summaries are derived directly from file contents.
  • Capability inventory: The skill executes Python scripts and writes data directly into the user's Obsidian Vault file structure.
  • Sanitization: Guidelines in references/safety-rules.md focus on excluding sensitive credentials like API keys but do not specify measures to sanitize or ignore embedded natural language instructions within ingested files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 02:51 AM
Security Audit — agent-trust-hub — obsidian-wiki-ingest