python-sdk

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The doc includes code examples that hard-code an API key (e.g., client = inference(api_key="inf_your_key") and api_key="inf_..."), which encourages passing secret values verbatim in code/outputs and would require the LLM to handle or reproduce those secrets directly if provided.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The docs contain multiple high-risk patterns (direct eval of untrusted input, enabling remote code execution, automatic file upload/URI exposure, webhook tools with secrets, browser automation filling credentials, and tools that can delete files) which can be abused for remote code execution and data exfiltration if used with untrusted inputs or misconfigured approvals.

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

• Potentially malicious external URL detected (high risk: 0.70). The agent configuration includes a skill that references an external markdown URL ("https://example.com/skills/api-docs.md"), which is likely fetched at runtime and could inject instructions/context into the agent, so it poses a runtime trust/control risk.