recoup-internal-upwork

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md Step 5/Step 1 (job search) and Step 1 (capture job) describe scraping Upwork job postings/descriptions and client history into applications/.../JOB.md, which the workflow can ingest as free text from outsider-authored Upwork content.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill contains curl calls that, at runtime, call the Apify API to fetch actor build metadata and to start a remote actor run (e.g. https://api.apify.com/v2/acts/neatrat~upwork-job-scraper and https://api.apify.com/v2/acts/neatrat~upwork-job-scraper/runs), which executes remote scraper code and is relied on by the job-search step.