bep-best-engineering-practice

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process software changes and system architecture descriptions provided by the user, which constitutes a surface for indirect prompt injection. Ingestion points: System changes and architectural documents under audit. Boundary markers: Instructions in SKILL.md and references/core.md explicitly require preserving the read-only task mode unless implementation is separately authorized. Capability inventory: No executable tools, network operations, or file-write capabilities are defined in the skill configuration. Sanitization: The 'Handle Failure Explicitly' section in references/core.md recommends validating untrusted input at system boundaries.
  • [SAFE]: The guidelines in references/core.md actively promote security best practices, including the exclusion of credentials and sensitive data from source control, logs, and caches.
  • [NO_CODE]: The skill consists entirely of instructional markdown and configuration files without any executable scripts or binaries.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 09:04 PM
Security Audit — agent-trust-hub — bep-best-engineering-practice