communication-history

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow uses HubSpot CLI commands like hubspot activities list and hubspot activities calls transcript get, which ingest outsider-authored free text from HubSpot records (e.g., activity title/body and call transcript utterances) into the agent’s LLM context via the CLI output.