PPT Office Assistant
Paste the following prompt into your AI chat to install this skill:
Please install @user_6e58ae68/jkljlkkljljljk according to https://skillhub.cn/install/skillhub.md.
About this skill
Problem Framing
In multi-turn PPT-related office tasks, the usual friction is not asking the right question, but stale context leaking into the request, reusing an outdated sessionId, or shell escaping corrupting the JSON body. This skill separates each call into keeping the current question verbatim, compressing only still-valid history, and storing session state in files, so the remote agent receives a cleaner prompt.
How It Works
- On first use, generate a session suffix, compact the current conversation’s still-relevant facts, and append the user’s latest question unchanged.
- Write the
promptpayload to a local JSON file and submit it through the configured proxy client. - Read the top-level
sessionIdfrom the response and immediately persist it; later turns read it from the file instead of relying on memory. - If the response returns
SESSION_EXPIRED, clear the old session files and retry once with a full, freshly compacted context. - Query the SSE endpoint line by line, parse
data:JSON objects, and concatenate theirtextfields into the final answer.
Boundaries
The capability accepts text input only and does not parse images, attachments, or documents. The product identifier and timestamp in file paths isolate sessions, so files from other sessions must not be reused. The output is streamed text, not a binary PPT file.
Use Cases
- Refine PPT outline and titles via text-only multi-turn asks.
- Compact valid notes into context before drafting slide points.
- On session expiry, resend context once and read new answer.
- Parse SSE data lines and join text into a reply.
Best For
- Ops preparing leadership reviews who want outline and script.
- PMs drafting proposals who need compact background and points.
- Support staff turning FAQ workflows into plain text replies.
- Engineers integrating proxies needing stable session files.
Related Skills
Generate and edit .pptx decks with python-pptx, applying structured layouts, design rules, native charts, and visual QA to reduce template-like output.
Tencent Cloud Table Recognition V3 is an OCR skill for detecting and recognizing tables in images or PDFs, supporting various table types like linear and borderless tables, with Excel export.
The complete set of online document operation tools provided by Tencent Docs MCP, supporting creation, querying, and editing of smart docs, Excel, PPT, mind maps, and more.
Generate structured Chinese patent disclosure drafts in Word from existing documents or brief technical inputs, following Part A/B/C templates.