obsidian-arrow-maintenance

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.70). The required workflow runs pnpm skills:update / npx skills add kylebrodeur/obsidian-arrow-sandbox --all --yes, which installs outsider-authored skill content from a public repo into the project and can be ingested by the agent as LLM-readable skill prompts at runtime (indirect prompt-injection risk).

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The file includes a runtime install command that pulls agent skills from the GitHub repo "kylebrodeur/obsidian-arrow-sandbox" via "npx skills add kylebrodeur/obsidian-arrow-sandbox", which will fetch and install remote skill code that can directly control agent prompts/instructions.