DocumentationPricingLoginSign up

Getting Started

  • Introduction
  • Installation & Setup

Features

  • Modules
  • Platform
  • Recipes
  • Templates
    • Template Directory
    • Custom Templates

Cluster

  • Details & Setup
  • Tasks, Actions & Agents
  • Assets & Files

API Reference

  • Command-Line Interface
  • Recipe Development
  • Recipes API

Templates new

Templates let you reuse, share and adapt project workflows for Ellf. They’re typically a repo or directory containing the project setup and overview of plans, tasks, actions, agents and services to create, as well as the required assets and custom recipe code.

The ellf-template module makes it easy to import and export templates – just point it at a directory or repo link and let Ellf do the rest. If decisions or resources are needed to adapt the template to your use case, your coding agent will ask follow-up questions.

✳︎Claude Code: Importing a project template from GitHub
› /ellf-template

Set up this project using the sample data: https://github.com/explosion/ellf-templates/tree/main/github-docs-issues


Template directory

Click on a template to view more details and the files, or copy-paste the URL to import it via your coding assistant. You can also ask it to adapt the template to your use case or different data. Alternatively, the template documents all commands to set up the project manually if you prefer. To contribute a template you’ve built, get in touch!

Only use templates from trusted sources and review them before giving your coding agent access to run code and populate your Ellf account on your behalf.

Replacing an LLM Classifier with Jev

Evaluate moving an LLM-based classifier to Jev, which answers with calibrated probabilities instead of text, and carry the move out. Compares Jev with the LLM on past records, has a person annotate a blind sample of disagreements and random records in about an hour, uses the annotations to add rules and choose per-label thresholds, and exports a configuration to run Jev in production, with optional checks for drift.

by Explosion
  • Text Classification
  • Evaluation
2 tasks3 actions1 planTypeSafe

Evaluating an LLM Classifier

Measure a classifier that's built on an LLM prompt against people, without changing it. An action reads its logged decisions, sampling by label, or runs it over a random sample of texts, and saves its answers as one more annotator; people label the same texts blind after a short pilot, the built-in review settles every disagreement, and the annotation dashboard shows how often it agrees with the settled labels and which labels it confuses.

by Explosion
  • Evaluation
  • Text Classification
4 tasks2 actions1 service1 recipe1 planYour LLM provider

Helpdesk Routing with Jev

Find out whether Jev is good enough to route IT support tickets to the right technician by problem area. Settles the label scheme in a pilot with a Gemini second opinion, builds a human-labelled evaluation set with an agreement check and a blind slip check, scores Jev against a bar agreed in advance, and sets up daily routing with a monthly problem-area report.

by Explosion
  • Text Classification
  • Evaluation
9 tasks3 actions1 agent4 plansGoogle AI, TypeSafesample data

Support Ticket Insights

Extract products, features, error codes, versions and integrations from customer support tickets and route each ticket to the right queue. An LLM drafts the entity annotations, people curate them, a pretrained pipeline suggests the generic entity types, and the passes are adjudicated into a gold set that trains a custom spaCy NER pipeline and a separate routing classifier.

by Explosion
  • Information Extraction
  • Text Classification
  • LLM Distillation
  • Evaluation
8 tasks6 actions1 agent2 services4 plansGoogle AIsample data

GitHub Docs Issue Classifier

Train a binary text classifier that predicts whether a GitHub issue is about documentation. Annotate a single label with accept or reject on the included GitHub issues or your own, then train a spaCy pipeline on the answers.

by Explosion
  • Getting Started
  • Text Classification
1 task1 action1 plansample data

Food Ingredient Extraction

Train a named entity recognizer that finds food ingredients in cooking comments from Reddit or your own texts. Highlight ingredients with match patterns as suggestions, train a quick first model on CPU and a more accurate transformer-based pipeline on GPU, and optionally correct the first model's suggestions in a second round and check the annotations in a dashboard.

by Explosion
  • Getting Started
  • Information Extraction
  • Named Entity Recognition
2 tasks2 actions1 service1 planGPUsample data
Read more

Templates Repo

Collection of open-source templates

Template Module

Import and export project templates

Creating custom templates

Using the ellf-template command, you can also export any project in your Ellf account as a template – for example, to make it easy to repeat end-to-end workflows or share them with others. The project will be documented in a consistent format that can later be imported automatically or manually.

Each template will have a README.md with an overview of the resources like tasks or actions that should be created, assets to upload to the cluster, including sample data to get started, and what needs to be provided or configured by the user when importing a template. During import, the agent will first consult this file.

✳︎Claude Code: Exporting a template for a project
› /ellf-template

Export my support tickets project as a template to this directory and include some sample data.

Example: Project template directory structure
📂 template
├─ 📄 README.md            # template overview and instructions
├─ 📂 assets               # optional assets and sample data
│  ├─ 📁 input             # input data, e.g. data to annotate
│  ├─ 📄 instructions.md   # annotation instructions
│  ├─ 📄 labels.json       # label schemes
│  └─ 📄 …                 # any other assets, e.g. patterns or configs
├─ 📁 config               # any configuration settings, e.g. for Prodigy
├─ 📂 plans                # project plans to create
│  └─ 📄 project_plan.md   # individual plan (with placeholders for links)
├─ 📂 recipes              # custom recipe packages used by project
│  └─ 📁 recipe_one        # individual standalone recipe package
└─ 📁 …                    # any other relevant resources, scripts etc.

Note on links in project plans

By default, Ellf will also include your project plans in the template. If the plan contains cross-reference links to resources in the platform, like assets or jobs, those will be replaced by a placeholder like @@Assets/Support Tickets@@ (instead of the original UUIDs, which might not be valid when re-using the template). When importing a template, Ellf will upload and set up all resources in the new project and then update the placeholders so the links work as expected.

Yes, after creating a template, you should definitely review it, and you can optionally make edits to improve the descriptions and include more information. However, it’s recommended to keep the directory structure and file names intact to make it easier for the agent to later create a project from the template.

Templates are designed to be published as Git repositories. You can create one repo per template, or add multiple templates in separate directories like we do in ellf-templates.

For the optional data included with a template, the agent will recommend using a small sample, e.g. a few hundred examples only, which is usually fine to commit directly to the repo. If you want to distribute a template with larger datasets, you could provide URLs instead. When importing your template, the agent will then first download the data before publishing it as an asset to your cluster. Alternatively, you could also include a custom action recipe that runs on the cluster, downloads the data and automatically creates an asset, or even create a custom Asset subclass that directly loads from a remote source.

If you’ve created a template and want to contribute it to this directory, upload it to GitHub and email us with the link. We especially like featuring templates that cover a specific but common enough use case and can be adapted for different data sources and business problems.

Ellf also provides an export command that saves the state of your organization to a JSON file and optionally downloads any assets and data. This is useful to back up your data, or move projects to a new organization, and it includes the exact state and contents of your projects.

Templates don’t necessarily include everything that’s in your project and instead document the project setup and turn it into a reusable workflow. You should use them if you want to be able to re-run the project with different data or settings.

from the makers of spaCy and Prodigy

Navigation

  • Home
  • Documentation
  • Templatesnew

Platform

  • Pricing
  • Free Trial
  • Login

Use Cases

  • Prodigy
  • Universities
  • LLM Evaluation

Resources

  • Case Studies
  • Videos
  • Blog
Terms & ConditionsPrivacy PolicyImprint© 2026 Explosion