skills.rest
sergueik
sergueikCOMMUNITY
Explore all skills by this author →
GitHub

uipath-ixp

Manage UiPath IXP document extraction projects, labelling, prompts, and model versions via the uip ixp CLI.

9|6|Updated Aug 25, 2017
Search Tags:#cli#taxonomy#document extraction#model publishing#uipath#ixp#labelling
One-click install
npx skills add https://github.com/sergueik/springboot_study --skill uipath-ixp-sergueik
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: uipath-ixp
Source: https://github.com/sergueik/springboot_study/tree/main/basic-uipath-skills/.github/skills/uipath-ixp
Command: npx skills add https://github.com/sergueik/springboot_study --skill uipath-ixp-sergueik

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Building and maintaining a UiPath IXP (Intelligent eXtraction Platform) document extraction project involves many CLI operations — creating projects, uploading documents, reviewing predictions, tuning extraction prompts, and publishing model versions — and doing them wrong (e.g., blind-confirming predictions or misusing corrections) silently corrupts training data and metrics. ## Core Features & Use Cases - Project & Taxonomy Management: Create projects with autopilot taxonomy suggestion or an imported taxonomy file, author field groups, fields, and data types, and configure the extraction model and pre-processing. - Document Review & Labelling: Download documents, review IXP predictions field-by-field, confirm correct values, correct OCR-garbled text, mark genuinely missing fields, and handle repeatable groups per occurrence. - Prompt Optimization Loop: Diagnose low-F1 fields, rewrite extraction instructions, wait out retrains with bounded polling, and roll back regressed prompts selectively. - Model Lifecycle: Pull metrics, list model versions, publish, tag, untag, unpublish, and roll back versions. - Use Case: You receive a folder of vendor invoices and want a trained extraction model. The skill creates the IXP project, suggests a taxonomy, reviews predictions against each invoice, iterates on prompts until F1 targets are met, and publishes the model version. ## Quick Start Ask the assistant to list your IXP projects with uip ixp projects list, or to create a new IXP project from a folder of documents and review its predictions.

Frequently Asked Questions about uipath-ixp

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

FAQPage Schema
How do I create a UiPath IXP project from documents?▼

Run uip ixp projects create with a name and a folder path containing supported files (PDF, PNG, JPG, TIFF, BMP, GIF). The command uploads the documents and suggests a taxonomy automatically; use --skip-taxonomy plus import-taxonomy if you already have a taxonomy file.

How do I review and confirm IXP extraction predictions?▼

Get predictions with uip ixp labellings get-predictions, download and read each document, then confirm correct fields with labellings confirm --fields. Confirm at field level after reviewing values against the document; never confirm all documents without review since F1 only measures agreement with confirmed labels.

When should I use --corrections in IXP labelling?▼

Use --corrections only for OCR-garbled values where the prediction found the right field and location but misread characters, such as MSIÓÓÓ601020 instead of MSI0601020. Never use it to flip wrong predictions, reformat normalized dates or amounts, or fix inferred values — those fields stay unannotated.

How do I improve low F1 scores in an IXP project?▼

Rewrite the per-field or per-group extraction instructions using fields update-prompts or groups update-prompts with location hints and real examples, then wait for the retrain and re-read metrics. Roll back instructions selectively for any field whose F1 drops by more than 0.1.

Can this skill deploy an IXP model to an Orchestrator folder?▼

No. The skill can publish, tag, unpublish, and roll back model versions with uip ixp projects publish, but binding a model to a folder or environment is a product-side step done in the UiPath UI with no CLI equivalent.

Why did occurrence indices change after confirming a line item?▼

The server returns annotation-prediction matched pairs first, so confirming one row of a repeatable group moves it to Occurrence 0 and renumbers the rest. Re-run get-predictions after every per-occurrence write and target rows by their values, or batch all confirmations into one --updates call.

Related Skills

Explore More Software Engineering Skills→
kofttlcckofttlcc
14

invoice-organizer

Extract vendor, date, and amount from PDFs, JPGs, and PNGs to rename and sort invoices.

Community
Intermediate
UiPathUiPath
141

uipath-maestro-flow

Coordinate UiPath Flow projects with the Maestro CLI for planning, validation, and deployment.

Official
Advanced
viinearaujoviinearaujo
1

uipath-ixp

Manage UiPath Intelligent eXtraction Platform project lifecycles via CLI.

Community
Advanced
ChatGPT-KBChatGPT-KB

replicate-issue

Replicate and validate GitHub issues against Archon using browser automation.

Official
Advanced