improve-architecture

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The agent uses operating system commands like open, xdg-open, or start to automatically display the generated HTML report to the user.\n- [DATA_EXPOSURE]: The skill performs extensive reading of codebase files and architectural documentation to identify technical debt and structural friction.\n- [INDIRECT_PROMPT_INJECTION]: By processing code from the project, the skill creates a surface where malicious instructions embedded in the codebase could influence the agent.\n
  • Ingestion points: Project source code, CONTEXT.md, and architectural decision records.\n
  • Boundary markers: The skill does not define specific delimiters to isolate ingested code content from agent instructions.\n
  • Capability inventory: The ability to write HTML files to the system temporary directory and execute shell commands to open them.\n
  • Sanitization: The skill explicitly forbids external scripts, stylesheets, or CDNs in the generated output, which prevents XSS or exfiltration via the user's browser.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 02:10 PM
Security Audit — agent-trust-hub — improve-architecture