tiangong-kb-sci-search

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly allows/encourages passing api_key / sci_api_key as explicit JSON fields that are forwarded as CLI flags, which requires the agent to include secret values verbatim in generated commands/outputs.

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/sci_search.sh, outsider-provided JSON_INPUT (e.g., query, input, claim, and inline filter/datefilter) is parsed and passed directly to npx @tiangong-ai/cli@latest research search via --query or --input, so the runtime ingests free text submitted by the user without selecting a specific pre-authored item first.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill defaults to running remote code at runtime via "npx @tiangong-ai/cli@latest", which fetches and executes a remote npm package (external code) that the wrapper relies on.