mcp_ocr
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| PADDLEOCR_VL_BIN | No | Path to paddleocr-vl CLI binary | paddleocr-vl |
| PADDLEOCR_VL_TIMEOUT | No | Timeout in seconds for PaddleOCR-VL | 120 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| ocr_text_visionA | Extract text from an image. Returns full text, per-line confidence, and bounding boxes. Backend: vision. Apple Vision OCR — fast on-device GPU/ANE inference (macOS 10.15+). Best for CJK + major European languages. Zero install on macOS. |
| ocr_layout_visionA | Extract text with layout analysis. Returns blocks with bounding boxes. Backend: vision. Apple Vision OCR — fast on-device GPU/ANE inference (macOS 10.15+). Best for CJK + major European languages. Zero install on macOS. |
| ocr_table_visionA | Extract tables from an image. Returns structured CSV/JSON. Backend: vision. Apple Vision OCR — fast on-device GPU/ANE inference (macOS 10.15+). Best for CJK + major European languages. Zero install on macOS. |
| ocr_batch_visionA | OCR multiple images at once. Returns consolidated results. Backend: vision. Apple Vision OCR — fast on-device GPU/ANE inference (macOS 10.15+). Best for CJK + major European languages. Zero install on macOS. |
| ocr_text_paddleA | Extract text from an image. Returns full text, per-line confidence, and bounding boxes. Backend: paddle. PaddleOCR CPU — cross-platform, mature model. Wider language support than Vision, but slower. Requires PaddleOCR Python package. |
| ocr_layout_paddleA | Extract text with layout analysis. Returns blocks with bounding boxes. Backend: paddle. PaddleOCR CPU — cross-platform, mature model. Wider language support than Vision, but slower. Requires PaddleOCR Python package. |
| ocr_table_paddleA | Extract tables from an image. Returns structured CSV/JSON. Backend: paddle. PaddleOCR CPU — cross-platform, mature model. Wider language support than Vision, but slower. Requires PaddleOCR Python package. |
| ocr_batch_paddleA | OCR multiple images at once. Returns consolidated results. Backend: paddle. PaddleOCR CPU — cross-platform, mature model. Wider language support than Vision, but slower. Requires PaddleOCR Python package. |
| ocr_text_paddleocr_vlA | Extract text from an image. Returns full text, per-line confidence, and bounding boxes. Backend: paddleocr_vl. PaddleOCR-VL — 0.9B vision-language model on Apple Silicon (M1+). Most accurate, 109 languages, supports tables/formulas/charts. Requires paddleocr-vl Swift CLI. |
| ocr_layout_paddleocr_vlA | Extract text with layout analysis. Returns blocks with bounding boxes. Backend: paddleocr_vl. PaddleOCR-VL — 0.9B vision-language model on Apple Silicon (M1+). Most accurate, 109 languages, supports tables/formulas/charts. Requires paddleocr-vl Swift CLI. |
| ocr_table_paddleocr_vlB | Extract tables from an image. Returns structured CSV/JSON. Backend: paddleocr_vl. PaddleOCR-VL — 0.9B vision-language model on Apple Silicon (M1+). Most accurate, 109 languages, supports tables/formulas/charts. Requires paddleocr-vl Swift CLI. |
| ocr_formula_paddleocr_vlB | Extract mathematical formulas from an image. Returns LaTeX. Backend: paddleocr_vl. PaddleOCR-VL — 0.9B vision-language model on Apple Silicon (M1+). Most accurate, 109 languages, supports tables/formulas/charts. Requires paddleocr-vl Swift CLI. |
| ocr_chart_paddleocr_vlA | Extract chart data from an image. Returns structured data. Backend: paddleocr_vl. PaddleOCR-VL — 0.9B vision-language model on Apple Silicon (M1+). Most accurate, 109 languages, supports tables/formulas/charts. Requires paddleocr-vl Swift CLI. |
| ocr_batch_paddleocr_vlA | OCR multiple images at once. Returns consolidated results. Backend: paddleocr_vl. PaddleOCR-VL — 0.9B vision-language model on Apple Silicon (M1+). Most accurate, 109 languages, supports tables/formulas/charts. Requires paddleocr-vl Swift CLI. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 14 tools
Each tool is uniquely identified by a combination of task (text, layout, table, batch, formula, chart) and backend (vision, paddle, paddleocr_vl). The backend distinctions are clearly described, preventing confusion between same-task tools. The two extra tasks (formula, chart) are only available for the paddleocr_vl backend, adding no ambiguity.
All tools follow a strict `ocr_<task>_<backend>` snake_case pattern. Every task and backend uses consistent vocabulary, making the naming scheme highly predictable and easy to navigate.
14 tools is well within the ideal range for a multi-backend OCR server. The count reflects three backends with shared core tasks plus two backend-specific extras, each earning its place without bloat.
The tool surface covers the primary OCR workflows: text extraction, layout analysis, table extraction, and batch processing across backends. The paddleocr_vl backend adds specialized formula and chart extraction, filling the expected niches. No obvious dead ends or missing critical operations.