security-health-inline

Pass

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands for environment management and the orchestration lifecycle. These include directory creation (mkdir), running build and type-checking scripts (pnpm build, pnpm type-check), and standard git operations for committing and pushing remediation results.\n- [EXTERNAL_DOWNLOADS]: The skill uses the bd (Beads) CLI tool for issue tracking and wisp management. These commands perform network operations to synchronize state and manage security issues on the Beads platform, which is integrated into the workflow.\n- [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection because it processes subagent-generated reports containing codebase data and passes them to other subagents.\n
  • Ingestion points: The orchestrator reads security-scan-report.md (which contains output from a codebase scan) in Phases 2, 5, and 6.\n
  • Boundary markers: Absent. There are no explicit instructions or delimiters used to ensure the vulnerability-fixer subagent ignores instructions that might be embedded in the report content.\n
  • Capability inventory: The vulnerability-fixer subagent has the capability to modify local files. The orchestrator has the capability to execute shell commands and perform git network operations.\n
  • Sanitization: No explicit sanitization or content validation of the security-scan-report.md is performed before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 18, 2026, 01:47 PM
Security Audit — agent-trust-hub — security-health-inline