search1api

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly fetches and ingests open web and social content (e.g., SKILL.md and reference/examples.md show s1 crawl and searches/services including reddit, github, x, YouTube, public websites) and then instructs the agent to read, summarize, and synthesize that untrusted user-generated content as part of its workflow, which can materially influence subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime "s1 crawl " command fetches arbitrary external pages (e.g., https://example.com/blog-post) and injects that fetched content into the agent's context for summarization, which directly controls prompts and thus poses a high-risk runtime dependency.