yichen-summary
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
bashtool to executemkdirfor directory management andcatfor writing conversation summaries to the local file system. These operations are essential for the skill's primary purpose of saving notes to an Obsidian vault. - [PROMPT_INJECTION]: The skill processes untrusted dialogue data and incorporates it into shell commands, creating an indirect prompt injection surface.
- [PROMPT_INJECTION]: (1) Ingestion points: Current user conversation history used to generate the summary (SKILL.md).
- [PROMPT_INJECTION]: (2) Boundary markers: The instructions explicitly use quoted heredocs (
<< 'EOF') in generated shell scripts to prevent the interpretation or expansion of dialogue content by the shell. - [PROMPT_INJECTION]: (3) Capability inventory: The skill utilizes
bashtool for file creation and directory management (SKILL.md). - [PROMPT_INJECTION]: (4) Sanitization: The skill provides specific rules for the agent to sanitize filenames, avoiding path traversal characters and shell metacharacters.
Audit Metadata