meeting-prep-cc
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes common shell commands such as grep, find, and xargs to search for strings related to meeting participants within a local Obsidian vault. These operations are restricted to the path defined in the VAULT_PATH environment variable or the default /root/obsidian-vault directory.
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from the local vault, which could potentially contain untrusted instructions. 1. Ingestion points: Vault files searched and read in SKILL.md (Steps 1, 2, and 3). 2. Boundary markers: The output uses markdown headers to structure the brief but does not implement specific delimiters or instructions to ignore embedded commands. 3. Capability inventory: The skill can read local files and write a markdown brief to the local disk; no network or administrative capabilities are used. 4. Sanitization: None; file contents are directly interpolated into the output.
Audit Metadata