codebase-overview

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill executes an external command /sourceatlas:overview. The origin, safety, and behavior of the 'sourceatlas' tool are not verified or defined within the skill repository, representing an unvetted dependency.
  • [CREDENTIALS_UNSAFE] (MEDIUM): The instructions explicitly direct the tool to scan 'configs' and 'high-entropy files'. These file types are high-probability locations for hardcoded API keys, database connection strings, and other secrets. Reading these into the model's context increases the risk of credential exposure.
  • [PROMPT_INJECTION] (MEDIUM): This skill is susceptible to Indirect Prompt Injection (Category 8). It processes untrusted external data (READMEs and config files from a codebase). Malicious instructions embedded in these files (e.g., hidden in markdown comments or config values) could be interpreted as commands by the AI agent during the analysis process.
  • Ingestion points: Local codebase files, specifically READMEs and configs.
  • Boundary markers: None identified. There are no instructions to ignore embedded commands within the files being scanned.
  • Capability inventory: Execution of external command /sourceatlas:overview.
  • Sanitization: None identified.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 07:20 AM