pp-obsidian

Warn

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download and install binary tools from the author's GitHub repository and the npm registry.
  • [REMOTE_CODE_EXECUTION]: Directs the agent to execute code from remote sources using go install and npx commands to set up the environment.
  • [COMMAND_EXECUTION]: Invokes the local obsidian-pp-cli binary to perform vault operations, including the ability to read the full raw contents of local markdown files into the agent's context.
  • [DATA_EXFILTRATION]: Documents a --deliver webhook:<url> feature that allows sending command results—including sensitive note data—directly to a remote HTTP endpoint.
  • [DATA_EXFILTRATION]: Includes a feedback mechanism that can be configured via the OBSIDIAN_FEEDBACK_ENDPOINT environment variable to transmit local data to a remote server.
  • [PROMPT_INJECTION]: Displays a surface for indirect prompt injection. Ingestion points: Raw markdown files from an Obsidian vault read via the notes and live-search commands. Boundary markers: Absent. Capability inventory: Shell command execution and network delivery via webhooks. Sanitization: Absent; content is processed in its raw form.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 18, 2026, 05:04 AM
Security Audit — agent-trust-hub — pp-obsidian