forked from EduCraft/curriculum-project-hub
hongjr03
db49a0d23d
Teachers convert many PDFs in one tool call with bounded Docmind concurrency. Each item keeps its own output_dir/document.md and UsageFact; failures are per-file. Mirror role skills to .cph/runtime-skills and CPH_RUNTIME_SKILLS_DIR so agents can Read SKILL.md instead of dead .claude/sandbox stubs.
97 lines
3.3 KiB
Markdown
97 lines
3.3 KiB
Markdown
---
|
|
name: pdf-to-md
|
|
description: >
|
|
Convert PDF documents to Markdown bundles using the convert_pdf_to_md tool.
|
|
Handles single or multiple PDFs (concurrent batch), Feishu attachments, and
|
|
local workspace files. Produces high-quality Markdown with LaTeX formulas
|
|
and extracted images.
|
|
---
|
|
|
|
# PDF to Markdown Conversion
|
|
|
|
## When to use
|
|
|
|
Use this skill when the user asks to convert a PDF (or several PDFs) to
|
|
Markdown, extract text from a PDF, or turn PDF documents into an editable
|
|
format.
|
|
|
|
## How it works
|
|
|
|
The `convert_pdf_to_md` tool (provided by the in-process `cph_hub` MCP server)
|
|
calls Alibaba Cloud Document Mind to parse each PDF. It:
|
|
|
|
- Extracts text in reading order (handles multi-column, scanned, and
|
|
multi-language documents)
|
|
- Converts mathematical formulas to **LaTeX** (`$...$` inline, `$$...$$` block)
|
|
- Extracts tables as Markdown tables
|
|
- Downloads embedded images into the output directory
|
|
- Writes a single `document.md` file plus image files **per** `output_dir`
|
|
|
|
There is **no** workspace `.mcp.json` source file. MCP tools are injected by
|
|
Hub at run start. Do not look for MCP or skill source under workspace
|
|
`.claude/` — those paths are sandbox stubs (often character devices) and are
|
|
not readable constitution.
|
|
|
|
## Where this skill text lives
|
|
|
|
Prefer the Skill tool when the runtime offers it. If you need to re-read these
|
|
instructions with Read:
|
|
|
|
- Workspace copy (always under cwd): `.cph/runtime-skills/pdf-to-md/SKILL.md`
|
|
- Absolute path env: `$CPH_RUNTIME_SKILLS_DIR/pdf-to-md/SKILL.md`
|
|
|
|
## Workflow
|
|
|
|
### One PDF from a Feishu message
|
|
|
|
1. Use `feishu_read_context` to find the `file_key` of the PDF attachment.
|
|
2. Use `feishu_download_resource` to download it into the workspace.
|
|
3. Use `convert_pdf_to_md` with `input_path` + `output_dir`.
|
|
|
|
### PDF already in the workspace
|
|
|
|
1. Use `convert_pdf_to_md` with `input_path` and `output_dir`.
|
|
|
|
### Multiple PDFs (concurrent)
|
|
|
|
1. Download or locate every PDF in the workspace first.
|
|
2. Call **`convert_pdf_to_md` once** with:
|
|
|
|
```json
|
|
{
|
|
"items": [
|
|
{ "input_path": "sources/a.pdf", "output_dir": "md/a" },
|
|
{ "input_path": "sources/b.pdf", "output_dir": "md/b" }
|
|
]
|
|
}
|
|
```
|
|
|
|
3. Hub submits Docmind jobs with bounded concurrency (default 3, max 8;
|
|
optional `concurrency` argument). Prefer this over N sequential tool calls.
|
|
4. **Each item must use a distinct `output_dir`** — the tool always writes
|
|
`document.md` inside that directory; shared dirs overwrite each other.
|
|
5. Partial failure returns per-file OK/FAIL lines; re-run only failed items.
|
|
|
|
## Important rules
|
|
|
|
- **Always** use `convert_pdf_to_md` for PDF→Markdown. Do NOT attempt to parse
|
|
PDFs yourself with Read, Bash, Python, or any other method.
|
|
- If `convert_pdf_to_md` fails because no capability connection is configured,
|
|
tell the user to ask their organization admin to configure the Aliyun
|
|
docmind credential in the admin web UI (组织后台 → 能力).
|
|
- The output directory will be created if it does not exist.
|
|
- After conversion, use `send_file` to send generated markdown (or a zip you
|
|
assemble) back to the user if they requested delivery.
|
|
|
|
## Output
|
|
|
|
Per `output_dir`:
|
|
|
|
- `document.md` — the main markdown file
|
|
- `*.jpg` / `*.png` — extracted images, referenced from the markdown
|
|
|
|
## Cost
|
|
|
|
Billed per page (0.04 CNY/page ≈ $0.0056/page for enhanced formula mode).
|
|
Each successful file records its own usage fact on the run ledger.
|