Create Skill

Warn

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill accepts arbitrary repository URLs (repo_link) and reference URLs (references) to fetch content from external, untrusted sources.
  • [COMMAND_EXECUTION]: The skill clones external repositories based on user input, which typically involves executing system-level commands such as git clone.
  • [PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection as it ingests data from external repositories or documents and uses that content to generate new agent instructions (SKILL.md). * Ingestion points: repo_link and references parameters in SKILL.md. * Boundary markers: No delimiters or safety instructions are present in the skill.global.md template to prevent the agent from obeying instructions embedded in the source material. * Capability inventory: The skill possesses the ability to write to the local file system (in .agents/skills/) and access the network. * Sanitization: There is no evidence of content sanitization or validation for the data retrieved from external sources.
  • [REMOTE_CODE_EXECUTION]: The requirement to perform a 'full project analysis' after cloning a repository suggests that the agent may interpret or execute code found within the untrusted repository to understand its structure and functionality.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 28, 2026, 12:50 PM