the-account-blueprint
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites via
WebSearchandWebFetch, as well as user-pasted content. - Ingestion points: Web search results, company websites, job postings, and user-provided contact lists or CRM exports.
- Boundary markers: The skill includes a dedicated reference file (
references/agent-security.md) with explicit instructions to treat all retrieved content as data, not instructions. It mandates quoting injection attempts rather than summarizing them. - Capability inventory: The skill uses web searching tools and can write to the
.agents/directory for persistence. - Sanitization: The instructions explicitly require HTML-escaping for any interpolated values and forbid the use of dangerous functions like
innerHTMLwhen emitting code. - [CREDENTIALS_SAFE]: The skill proactively warns against credential exposure. The
references/agent-security.mdfile contains rules to redact tokens, avoid persisting secrets in files, and never ask for API keys. - [SAFE]: The skill contains comprehensive instructions to avoid using protected personal categories (race, religion, age, etc.) in any scoring or routing logic, ensuring compliance with ethical and legal standards.
Audit Metadata