feedback-ls

Send document pages to Label Studio for human feedback on transcriptions and segmentations.

17|2|Updated Jun 5, 2025
One-click install
npx skills add https://github.com/AI-Riksarkivet/ra-mcp --skill feedback-ls
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: feedback-ls
Source: https://github.com/AI-Riksarkivet/ra-mcp/tree/main/plugins/ra-mcp-tools/skills/feedback-ls
Command: npx skills add https://github.com/AI-Riksarkivet/ra-mcp --skill feedback-ls

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill streamlines the process of sending document pages to Label Studio for human feedback on AI-generated transcriptions and segmentations, ensuring higher quality and accuracy.

Core Features & Use Cases

  • Human Feedback Workflow: Facilitates sending document pages to Label Studio for annotation and review.
  • Two Modes: Supports importing pages with pre-annotations (using ALTO XML) or as blank image tasks for annotation from scratch.
  • Use Case: A user identifies several pages with transcription errors and asks to have them corrected. This Skill can send those specific pages to Label Studio, either with the existing AI transcription for correction or as a blank canvas for manual annotation.

Quick Start

Send the document page with the image URL 'https://lbiiif.riksarkivet.se/arkis!30002056_00004/full/max/0/default.jpg' and ALTO URL 'https://sok.riksarkivet.se/dokument/alto/SE_RA_30002056_00004' for transcription feedback.

Frequently Asked Questions about feedback-ls

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I send document pages to Label Studio for human review of AI transcriptions?

You can send document pages for human review by providing an image URL and an ALTO XML URL. This creates a Label Studio task with pre-annotations, allowing reviewers to correct AI transcriptions and segmentations efficiently.

Can I send image-only document pages to Label Studio for annotation from scratch?

Yes, you can send image-only document pages to Label Studio for annotation from scratch. Providing just the image URL creates a blank canvas task, allowing human reviewers to manually annotate the document page without any pre-existing AI transcriptions.

What is the ALTO XML format used for in transcription feedback workflows?

ALTO XML stores pre-annotated transcription and segmentation data for document pages. Sending this format to Label Studio overlays existing AI-generated text on the image, enabling human reviewers to verify and correct the output efficiently.

Does this workflow support correcting segmentation boundaries alongside transcription text?

Yes, the workflow supports correcting both segmentation boundaries and transcription text. When document pages are sent to Label Studio using ALTO XML, reviewers can adjust layout segmentation and corresponding text during the quality assurance process.

When should I use pre-annotated tasks instead of blank image tasks for document review?

Use pre-annotated tasks with ALTO XML when you have existing AI transcriptions needing human correction. Use blank image tasks when you want reviewers to manually transcribe and segment document pages from scratch without initial AI-generated content.