sdd-archive

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs filesystem operations using native shell commands (cp, mv, git mv, diff, mktemp). This is an intentional design choice to maintain byte-identity and prevent model-induced truncation during archival.- [DYNAMIC_EXECUTION]: Shell commands are assembled by interpolating variables like {change-name} and {domain}. This pattern creates a surface for command injection if the input variables are maliciously crafted and not sanitized before execution.- [INDIRECT_PROMPT_INJECTION]: The skill processes content from various artifacts which may contain untrusted data.
  • Ingestion points: Reads implementation tasks, verification reports, and delta specs from files like tasks.md and verify-report.
  • Boundary markers: Utilizes allowedEditRoots and actionContext checks to limit the scope of filesystem operations.
  • Capability inventory: Executes shell commands for copying, moving, and comparing directories and files.
  • Sanitization: Does not specify explicit escaping or sanitization for variables interpolated into shell commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 03:16 PM
Security Audit — agent-trust-hub — sdd-archive