annotate — Claude Code command for PDF markup
This Claude Code command annotates PDFs with an agent. It proposes changes section by section, then shows each batch for review.
Apache-2.0 ★ 36,654 updated 2026-10-24
/plugin marketplace add anthropics/knowledge-work-plugins
/plugin install pdf-viewer@knowledge-work-pluginsWhat annotate does
The command guides PDF annotation one section at a time. The agent reads page text, proposes markup, and applies it after approval; explicit instructions skip the proposal. Available edits include text highlights, underlines, strikethroughs, sticky notes, on-page comments, shapes, and labeled stamps.
After each batch, the agent shows a page screenshot and waits for feedback before moving on. It uses display_pdf and interact to open and edit the document, then checks the result with get_screenshot. The annotated PDF can be downloaded from the viewer toolbar.
What it helps with
- highlight a passage in a PDF
- place a note beside a contract clause
- stamp pages as CONFIDENTIAL
- review annotation changes in a screenshot
Example prompts for the agent
- Annotate
contract.pdf: highlight the termination clause and add a note about the 30-day window. - Stamp every page of
report.pdfwithCONFIDENTIAL. - Underline paragraph three in
brief.pdfand show me the resulting screenshot.
How to install annotate
1Plugin marketplace
/plugin marketplace add anthropics/knowledge-work-plugins/plugin install pdf-viewer@knowledge-work-plugins2Manually: put the files in place
.claude/commands/annotate.mdCall it with /annotate. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/annotate.mdCall it with /prompts:annotate. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/annotate.prompt.mdManually: put the files in place
.gemini/commands/annotate.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about annotate
How do I install annotate in Claude Code?
Run /plugin marketplace add anthropics/knowledge-work-plugins, then /plugin install pdf-viewer@knowledge-work-plugins. The command file is .claude/commands/annotate.md.
What does annotate do?
It adds markup to PDFs, including text highlights, comments, shapes, and stamps. The agent shows a screenshot after each batch so the changes can be reviewed.
What does annotate need to work?
It uses display_pdf and interact to open, read, and edit the PDF, plus get_screenshot to show the result.
Source text
annotate.md · Apache-2.0
---
description: Collaboratively annotate a PDF — propose markup, review together, iterate
argument-hint: "[path-or-url]"
---
> If you need to check which tools are connected, see [CONNECTORS.md](../CONNECTORS.md).
# Annotate PDF
Walk through a document with the user, proposing and applying
annotations section by section. The user reviews each batch in the
live viewer before you continue.
## Workflow (AI-driven default)
1. **Open** — `display_pdf` (or use existing `viewUUID` if already open)
2. **Understand** — `interact` → `get_text` on the first page range
(≤20 pages) to read content
3. **Propose** — describe to the user what you plan to annotate:
> "I'll highlight the termination clause on page 2, add a note
> 'Review 30-day window' next to it, and stamp page 1 as DRAFT.
> Sound good?"
4. **Apply** — on approval, `interact` with batched commands:
`add_annotations` + `get_screenshot` of the affected page
5. **Review** — show the screenshot, ask for edits
6. **Iterate** — move to the next section, repeat
7. **Finish** — remind the user they can download the annotated PDF
from the viewer toolbar
## Manual mode
If the user gives explicit instructions ("highlight paragraph 3",
"stamp CONFIDENTIAL on every page"), skip the proposal step and
execute directly. Still confirm with a screenshot.
## Annotation types available
- **Text markup:** `highlight_text` (auto-finds text — preferred),
highlight, underline, strikethrough
- **Comments:** note (sticky), freetext (visible on page)
- **Shapes:** rectangle, circle, line
- **Stamps:** any label — APPROVED, DRAFT, CONFIDENTIAL, REVIEWED
- **Images:** signatures, initials, logos (see `/pdf-viewer:sign`)
## Tips
- Prefer `highlight_text` over manual `rects` for text — it finds
coordinates automatically
- Batch related annotations in one `interact` call
- End each batch with `get_screenshot` so the user sees the result
- Keep proposals small (3–5 annotations per batch) so review is easy
© Anthropic · Apache-2.0 · anthropics/knowledge-work-plugins