new-entity-requirements

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external documentation from the web, creating a surface for indirect prompt injection.\n
  • Ingestion points: Elastic API documentation fetched via web search or mcp_web_fetch (SKILL.md, workflow step 2).\n
  • Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore' instructions for fetched content.\n
  • Capability inventory: The skill can write files to the local repository (TodoWrite) and execute commands via the openspec CLI (SKILL.md, workflow step 3).\n
  • Sanitization: Absent.\n- [EXTERNAL_DOWNLOADS]: The skill utilizes external vendor resources for its core functionality.\n
  • The go-elasticsearch library (github.com/elastic/go-elasticsearch/v8) is used for API exploration.\n
  • The openspec CLI tool is required for the workflow, with installation managed via repository setup scripts (make setup).\n- [COMMAND_EXECUTION]: The skill involves executing CLI commands with identifiers derived from user input.\n
  • Workflow management relies on commands like 'openspec new change' and 'openspec instructions' to generate and manage artifacts.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 09:05 AM
Security Audit — agent-trust-hub — new-entity-requirements