name: nano-pdf
description: Edit PDFs with natural-language instructions using the nano-pdf CLI.
homepage: https://pypi.org/project/nano-pdf/
metadata:
{
"openclaw":
{
"emoji": "📄",
"requires": { "bins": ["nano-pdf"] },
"install":
[
{
"id": "uv",
"kind": "uv",
"package": "nano-pdf",
"bins": ["nano-pdf"],
"label": "Install nano-pdf (uv)",
},
],
},
}
Agent for applying natural-language edits to specific PDF pages using the nano-pdf CLI tool.
- Tooling: `nano-pdf` CLI.
- Execution Mode: Terminal execution.
- Capability Scope: Single-page targeted PDF editing using natural language prompts.
- MUST specify exactly one target PDF file, one page number, and one clear instruction string.
- Page numbers are either 0-based or 1-based depending on the tool's version/config.
- MUST handle off-by-one page index errors by retrying with the shifted page number if the first attempt edits the wrong page.
- NEVER assume the output is perfect; always prompt the user to sanity-check the result.
- AVOID running on encrypted or password-protected PDFs without first decrypting them if necessary.
1. Identify the target PDF file, the specific page number to edit, and the natural language instruction.
2. Execute the CLI command: `nano-pdf edit ""`
3. If the result is off by one page, increment or decrement the `` and re-run.
4. Provide the path to the edited PDF file to the user.
- Provide the exact file path of the edited PDF.
- Remind the user to manually verify the output PDF for formatting and correctness.
- Do not attempt to read the PDF back into context to verify.
By x10aix schedule Updated 5/21/2026