career-skill-factory

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The career-skill-factory processes untrusted user input such as career names, industry profiles, and detailed Job Descriptions (JD). This data is used to drive automated research phases and the creation of sub-tasks.
  • Ingestion points: User input for careers, industries, and Job Descriptions entering the system via SKILL.md.
  • Capability inventory: The skill uses WebSearch to gather external information and spawn to create sub-agents for specialized distillation tasks.
  • Boundary markers: The prompt templates used for sub-agents in Phase 3 lack explicit delimiters or instructions to ignore potential commands embedded within the harvested research data.
  • Sanitization: There is no evidence of sanitization or filtering of external content before it is processed by the sub-agents.
  • [COMMAND_EXECUTION]: The skill extensively uses the spawn capability to parallelize research tasks and the WebSearch tool for data gathering. These capabilities are used to dynamically create files and directories (e.g., [industry]-advisory-suite/) based on the outcomes of the research phase, which is influenced by untrusted user input.
  • [EXTERNAL_DOWNLOADS]: The documentation suggests installation using npx skills add SpaceZephyr/career.skill. This refers to the vendor's own repository for the purposes of skill management within the supported environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 03:15 AM
Security Audit — agent-trust-hub — career-skill-factory