llm-wiki-obsidian-hardening
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a defensive framework for AI-assisted vault editing. It mandates safety gates such as verifying git status or backups before performing any write operations.
- [COMMAND_EXECUTION]: The skill recommends utilizing standard tools like
rg(ripgrep) or the Obsidian CLI for search tasks. These are documented as preferred alternatives to insecurely reading the entire vault content. - [INDIRECT_PROMPT_INJECTION]: The skill proactively addresses the risk of processing untrusted markdown content by providing specific procedures for implementing boundary markers (
AGENTS.md,CLAUDE.md) and directory isolation (raw/vswiki/) to prevent the agent from misinterpreting or overwriting human synthesis.
Audit Metadata