IMA OpenAPI Unified Access

Problem

Calling IMA OpenAPI often fails in two places. First, notes text writes can receive non-UTF-8 input from users, web fetches, files, or clipboard, producing irreversible garbled content. Second, the boundary between notes and knowledge-base is easy to misroute: appending to a note, adding a note to a knowledge base, and reading source content from a knowledge base each need different paths. The skill consolidates these operations behind one entry point and enforces validation and routing before execution.

How It Works

• Module routing: use notes for searching, browsing, creating, and appending notes; use knowledge-base for file uploads, link addition, knowledge-base search, and associating notes. Cross-module tasks require reading both SKILL.md files first.
• Credential and request flow: check environment variables or ~/.config/ima before API calls; send requests as HTTP POST + JSON Body to https://ima.qq.com.
• Encoding safeguards: validate content and title as UTF-8 before import_doc or append_doc; keep uploaded files as original binary content without renaming or re-encoding.
• Error handling: distinguish script-level errors -100 and update hints -200 from backend JSON code and msg, surfacing actionable failure messages.

Limits and Caveats

This skill fits Agent workflows that need IMA note and knowledge-base access. Video files, Bilibili/YouTube URLs, and file:// URLs are unsupported. In PowerShell 5.1, requests must use UTF-8 byte-array mode. If the target note is ambiguous, confirm whether to create a new note or append to an existing one.