polite-mcp-conventions-taudelta
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill outlines defensive behavioral guidelines for agents interacting with personal shopping and classifieds accounts. It explicitly mandates a confirmation step via
AskUserQuestionbefore the first mutating action in a session, which serves as a security safeguard against unintended or malicious account changes. - [INDIRECT_PROMPT_INJECTION]: The skill is intended to process external web data (e.g., product pages on OLX, eMAG, Temu). While this represents a surface for indirect prompt injection, the skill's core conventions—requiring user confirmation for mutations and batching decisions for user review—are effective mitigations that maintain human-in-the-loop control over the automation process.
- [COMMAND_EXECUTION]: The skill references writing findings into
PLAN.md. This is a documented, internal convention for project documentation and does not involve the execution of arbitrary or dangerous system commands.
Audit Metadata