Baitong Academic Article Search

Problem

When an agent calls an academic article search capability, running the search script directly can skip two prerequisite problems: an unusable local Python runtime and incomplete authorization for a protected Baitong service. This skill places environment detection, credential binding, and failure reporting before the business workflow, making service availability and output permission explicit gates instead of silent retries or unverified answers.

How It Works

• Runtime check first: locate a usable Python interpreter before running scripts. If detection fails, report with the expected template. If the user provides a full path, use it directly.
• Automatic authorization binding: run scripts/auth.py --ensure to check local cache and platform validation. If it returns AUTH_REQUIRED or CREDENTIALS_MISSING, proactively call connect_cloud_service to obtain the current user JWT, then run --bind-jwt and re-run --ensure.
• Protected content fetch: only after scripts/protected.py returns protected instructions should the retrieval workflow continue. Business reasoning should not happen before AUTH_OK.
• Traceable failures: distinguish network errors, remote authorization failures, and VIP expiration. Retry only within the stated limit, then stop and provide a clear message.

Boundaries

The skill requires a usable Python runtime, network access, Baitong platform authorization, and WorkBuddy login state. It should not expose AppKey or AppSecret in plaintext, and it should not bypass authorization to return search results. Search scope, fields, and platform policy are governed by the protected interface response.