jina-cli

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt includes examples that set API keys via the CLI (e.g., "jina config set key YOUR_API_KEY") and passes credentials on the command line (e.g., --cookie "session=abc123"), which requires embedding secret values verbatim in commands and outputs.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). While many listed links are benign content/read URLs (x.com posts, docs, jina.ai endpoints), the skill explicitly instructs piping a raw GitHub-hosted install.sh (raw.githubusercontent.com/geekjourneyx/...) to bash and references a proxy (http://proxy.com:8080), which is a high‑risk distribution pattern for arbitrary/remote code execution from an unverified repo and therefore suspicious.

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

• Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md explicitly instructs the tool to fetch and extract content from arbitrary public URLs (e.g., "jina read --url 'https://x.com/user/status/123'", batch URL files) and to auto-fetch top web search results, meaning it ingests untrusted, user-generated web content that can influence downstream decisions.

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

• Potentially malicious external URL detected (high risk: 0.90). The CLI uses the Jina Reader/Search APIs at runtime (JINA_API_BASE_URL=https://r.jina.ai/ and JINA_SEARCH_API_URL=https://s.jina.ai/) to fetch processed web content that is injected into the model context and can directly control prompts/output, so these external endpoints are required runtime dependencies that influence agent behavior.