Skills
Navigate
skills/vesviet/agent-skills/add-ui-component/Gen Agent Trust Hub

add-ui-component

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from structured specifications which could potentially contain malicious instructions.
  • Ingestion points: The skill reads from contracts/schemas/ui-component-spec.json and contracts/schemas/ux-flow-spec.json (SKILL.md).
  • Boundary markers: While schema validation is mentioned in guardrail ASI05, explicit prompt delimiters for content interpolated into the agent's context are not defined.
  • Capability inventory: The skill possesses significant capabilities including write_file, edit_file, and execute_command (via allowed-tools in SKILL.md).
  • Sanitization: Guardrail ASI05 specifies strict schema validation for user-supplied content to mitigate RCE risks.
  • [COMMAND_EXECUTION]: The skill utilizes the execute_command tool to run local scripts and development tools.
  • Evidence: The skill is configured to use execute_command for tasks like running tests and linters. The UPGRADE_PLAN_2026.md explicitly mentions running a local validation script: python3 core/scripts/validate-all.py.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 11:20 AM