paw-mkt-sales

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to fetch and scrape public, user-generated and third-party pages (via agent-browser, WebFetch/WebSearch) — see references/competitive-research.md and references/shared-patterns.md which call out opening G2, Google search, LinkedIn, competitor sites and using "agent-browser get text body" — and those live pages are read and used to drive competitive research and subsequent actions (battle cards, claims, collateral), so untrusted third‑party content is ingested and can materially influence behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill declares and uses the external agent-browser dependency installed from https://github.com/vercel-labs/agent-browser at runtime (via "npx skills add https://github.com/vercel-labs/agent-browser") and then uses agent-browser commands to fetch arbitrary live webpages (e.g., competitor sites, G2, LinkedIn) that are injected into the agent's context and thus directly control prompts/behavior.