spec-engine

Warn

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The file tests/run-validation.py uses the exec() function to dynamically execute the code contained within the local script scripts/validate-output.py during the test process.
  • Evidence: exec((SCRIPTS_DIR / "validate-output.py").read_text(encoding="utf-8"), validate_output) in tests/run-validation.py.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external sources, including user research, tickets, and normalized handoff JSON from Azure DevOps (ado-gateway), creating a potential vector for indirect prompt injection.
  • Ingestion points: SKILL.md (Workflow section), rules/handoff-consumption.md (Azure DevOps handoffs).
  • Boundary markers: The rules/handoff-consumption.md file specifies that the agent should reject malformed JSON, verify the producer field, and treat pr_comments as review context only rather than functional requirements.
  • Capability inventory: The skill uses the read_file tool to interact with the filesystem.
  • Sanitization: rules/requirement-quality.md provides strict checks, banning vague language (e.g., "should", "faster", "user-friendly") and requiring testable outcomes to minimize the impact of ambiguous or malicious inputs.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 7, 2026, 11:25 AM