Skills
skills/gooseworks-ai/goose-skills/customer-win-back-sequencer/Gen Agent Trust Hub

customer-win-back-sequencer

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted data from the web (search results, career pages, and review sites) and incorporating it into its reasoning logic and output generation for email sequences.
  • Ingestion points: Phase 1 uses web_search and fetch_webpage to retrieve external content from sites like LinkedIn, G2, and various news outlets.
  • Boundary markers: The skill instructions do not define delimiters or specific constraints to prevent the agent from following malicious instructions that might be embedded in the fetched external content.
  • Capability inventory: The agent has permissions to write reports to the file system and potentially trigger automated email campaigns via the cold-email-outreach tool.
  • Sanitization: There is no evidence of sanitization, filtering, or validation of the fetched external data before it is interpolated into the agent's prompts.
  • [EXTERNAL_DOWNLOADS]: Installs the skill and its dependencies (including linkedin-profile-post-scraper and review-site-scraper) via the npx goose-skills command.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 10:47 AM