team-boot
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and aggregate data from multiple local project files (
.adlc/memory/pdr/pdr.md,.adlc/memory/adr/adr.md,.adlc/memory/chdr.md, and context modules) and inject their contents directly into the system prompt via an automatic session hook. While the skill's own code is safe, this pattern creates a vector for Indirect Prompt Injection (Category 8c - Tool Output Poisoning) where a malicious actor could place adversarial instructions inside these markdown files (e.g., hidden HTML comments or deceptive metadata) to influence agent behavior at runtime. The risk is minimized as it requires local file access, but the ingestion of untrusted/external data into the prompt surface is present.
Audit Metadata