About this skill

What it solves

Calling the IMA OpenAPI can easily mix two paths: append_doc adds content to an existing note, import_doc creates a new note, while file uploads, link additions, and source retrieval belong to knowledge-base. Chinese text read from files, web pages, or sent through PowerShell 5.1 can also be silently converted to GBK, causing unrecoverable mojibake. This skill turns routing, credential checks, UTF-8 validation, and error-code handling into a consistent set of invocation constraints, forcing the agent to decide whether the target is note content, a knowledge-base item, or the source of a knowledge-base item.

How it works and where it stops

Routing: notes use notes; uploads, web links, and knowledge-base search use knowledge-base; cross-module tasks read both submodules before acting.
Validation: content and title must be valid UTF-8 before import_doc or append_doc; uploads remain binary, and title must match the original file_name.
Credentials and requests: environment variables or config files are checked, and requests are sent only to https://ima.qq.com as POST + JSON.
Errors: script-level codes include -100 and -200; business-level errors are returned in stdout with code != 0 and should surface msg directly.
Limits: video, Bilibili/YouTube URLs, and file:// URLs are unsupported; PowerShell 5.1 must send an explicit UTF-8 byte array to avoid silent GBK conversion.

Screenshots

Use Cases

Save a web article as a knowledge-base item, add its link, and prepare it for later retrieval before publishing support docs.
Append a drafted update to a specified existing note by searching the note first and calling the append flow with the right note ID.
Write Chinese note content from a web page or file, validating UTF-8 before import_doc or append_doc to avoid garbled text.
Inspect a note-type item in a knowledge base by reading its media info, then fetch the original note content for review.

Best For

Backend engineers integrating IMA automation: need correct module routing for notes and knowledge-base calls.
Ops engineers maintaining Chinese docs: need to write web or file content into notes without UTF-8 garbling.
Developers calling IMA APIs from PowerShell 5.1: need to avoid silent GBK conversion of request bodies.
Support engineers using knowledge-base retrieval: need to find items, inspect original content, and distinguish note append from KB add.