omc-doctor
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands (
ls,grep,mkdir,rm) andnode -einline scripts to perform system diagnostics, verify installation states, and fix environment issues. These actions are aligned with the skill's primary purpose as a maintenance utility and are restricted to the user's configuration directory (defaulting to~/.claude). - [EXTERNAL_DOWNLOADS]: The skill fetches markdown content from the author's GitHub repository (
raw.githubusercontent.com/Yeachan-Heo/oh-my-claudecode/main/docs/CLAUDE.md) to update the local documentation. This download targets a well-known service and the vendor's own infrastructure, posing no inherent risk as it does not execute remote code. - [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface that includes local settings files, markdown files, and content fetched from GitHub.
- Ingestion points: Reads
settings.json,.omc-config.json, andCLAUDE.mdfrom the local file system (SKILL.md Step 2, 4). - Boundary markers: The instructions mandate checking for specific markers (
<!-- OMC:START -->) and verify file integrity using SHA-256 fingerprints before performing modifications. - Capability inventory: Contains capabilities for file deletion (
rm -rf), file creation (mkdir,touch), and shell execution (node -e). - Sanitization: Employs strict verification logic, comparing script contents against hardcoded SHA-256 fingerprints and requiring explicit user confirmation before modifying or removing settings and scripts.
- [DYNAMIC_EXECUTION]: Employs
node -eto execute small JavaScript snippets for cross-platform path resolution, file system traversal, and version comparison. This is used as a utility for the diagnostic process rather than for generating arbitrary executable logic.
Audit Metadata