client-packet-engine

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface.\n
  • Ingestion points: The skill ingests untrusted data from user-provided URLs in SKILL.md during Phase 1 (Intelligence Gathering) and Phase 3 (Strategy Execution) via tools like review-scraper, web-archive-scraper, and luma-event-attendees.\n
  • Boundary markers: The instructions do not define any delimiters or explicit boundary markers to isolate external web content from the agent's reasoning or to prevent the agent from following instructions embedded in the target websites.\n
  • Capability inventory: The skill utilizes exploitable capabilities such as email-drafting (for outreach sequences) and content-asset-creator (for generating HTML assets), which are direct sinks for the ingested untrusted data.\n
  • Sanitization: No sanitization, filtering, or validation logic is specified to check the content of external websites before it is processed by generative sub-skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 05:18 PM