job-search

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and scrapes job postings from arbitrary public URLs (see Stage 1: scripts/fetch-jd.py / "apply.sh ") and performs daily automated scraping of public job boards (Stage 3, scan-config.json with Ashby/Greenhouse), and those untrusted job descriptions are fed into multiple agents (eligibility, cv-tailor, cover-letter, ats-checker, interview-research, Haiku) that make decisions or generate follow-up actions (e.g., aborting, re-tailoring, writing to Notion), so third-party content can materially influence the agent's behavior.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill fetches job posting URLs at runtime (e.g., https://jobs.example.com/senior-engineer and https://jobs.ashby.io/company/job-id) via scripts/fetch-jd.py and injects the fetched job description directly into agent prompts (eligibility, cv-tailor, cover-letter), so external content controls the agents' outputs.