kol-engager-icp
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from LinkedIn, including post text, comments, and profile headlines, to perform lead classification and enrichment.
- Ingestion points: Data is pulled from LinkedIn via Apify actors in scripts/kol_engager_icp.py (functions scrape_kol_posts, scrape_engagers, enrich_profiles).
- Boundary markers: There are no explicit delimiters or warnings to the agent to ignore instructions embedded in the scraped text (e.g., in LinkedIn comments).
- Capability inventory: The skill performs network requests to Apify APIs, writes CSV and JSON files to the local file system, and executes external Apify actors.
- Sanitization: The script performs basic data extraction and regex matching for classification, but does not sanitize the text for potential prompt injection vectors that could affect an agent reading the output.
- [EXTERNAL_DOWNLOADS]: The skill relies on external Apify actors to perform LinkedIn scraping.
- Evidence: References harvestapi/linkedin-profile-posts, harvestapi/linkedin-company-posts, and supreme_coder/linkedin-profile-scraper.
- Source: Data is fetched from api.apify.com using an API token provided by the user in a .env file.
- [COMMAND_EXECUTION]: The skill instructions provide shell commands for execution and installation.
- Evidence: SKILL.md includes commands for running the Python pipeline and installing the skill via npx.
Audit Metadata