mj-skill-creator

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill automates various shell commands to manage the development lifecycle, including git operations (status, add, commit, push) and the execution of local Python maintenance scripts located in ~/.codex/skills/.system/.
  • [EXTERNAL_DOWNLOADS]: It utilizes npx skills add to fetch and globally install skill packages from the vendor's own GitHub repository (makerjackie/skills). Use of the --yes flag automates this process by bypassing manual confirmation prompts.
  • [DATA_EXFILTRATION]: As part of its publishing workflow, the skill pushes locally generated or updated skill source code to a remote GitHub repository. This is the intended behavior for deploying the created skills.
  • [REMOTE_CODE_EXECUTION]: The skill executes code from a remote source by installing packages directly from GitHub into the agent's environment. This behavior is restricted to the vendor's repository and is consistent with the skill's primary purpose.
  • [DYNAMIC_EXECUTION]: The tool dynamically generates functional skill components, including SKILL.md and openai.yaml configuration files, based on user-provided descriptions of tasks and workflows.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 04:31 AM
Security Audit — agent-trust-hub — mj-skill-creator