datatailr/datatailr-skills

Use the Datatailr platform to build and deploy data processing workflows, long running services with endpoints, web apps and Excel add-ins.

Manage Datatailr cluster administration via CLI and Python SDK. Inspect and modify compute nodes (Node / dt node), platform settings (dt settings) including autoscaler capacity, idle time, drain/terminate, login, and ACLs, and cost/budgets (Cost, Budget / dt cost). Use when the user asks about admin settings, autoscaler, cluster capacity, nodes/instances, idle time, drain, terminate, platform settings, spend, cost view, or budgets on Datatailr / dt.

Use Datatailr blob storage to persist and retrieve files and data. Blob storage is a key-value store for arbitrary binary data accessible from any job. Use when the user asks to store data, save files, share data between jobs, read/write blobs, or use persistent storage on Datatailr / dt.

Customize the Docker image for a Datatailr job using build_script_pre and build_script_post. Covers installing OS packages (apt-get), compilers, system libraries, Cython / native extension compilation, and build script formatting. Use when the user's job needs system-level dependencies, native code compilation, or any image customization beyond pip packages.

Deploy a web app on the Datatailr platform using the Python SDK. Supports Streamlit, Dash, Flask, FastAPI, Panel, and standalone Notebook (Voila) apps. Use when the user asks to create and deploy an app, dashboard, web app, or notebook app on Datatailr / dt.

Deploy an Excel add-in on the Datatailr platform using the Python SDK. Excel add-ins expose Python functions as Excel worksheet functions. Use when the user asks to deploy an Excel add-in, expose Python functions to Excel, or create Excel UDFs on Datatailr / dt.

Deploy a long-running service on the Datatailr platform using the Python SDK. Services are background processes like API servers, message consumers, or workers that auto-restart on exit. Use when the user asks to deploy a service, API, or background worker on Datatailr, or to make existing service code run on Datatailr / dt.

Deploy a DAG workflow on the Datatailr platform using the Python SDK. Workflows are directed acyclic graphs of tasks that run in parallel when possible. Use when the user asks to deploy a workflow, data pipeline, batch job, or DAG on Datatailr / dt, or to make existing Python functions deployable as a workflow.

Run distributed scikit-learn / joblib workloads on the Datatailr platform. Parallelizes joblib.Parallel work (grid search, cross-validation, random forests) across a pool of worker containers with no manual setup. Use when the user asks to distribute, scale out, or speed up scikit-learn or joblib code on Datatailr / dt.

Monitor and troubleshoot Datatailr jobs using the CLI. Check job statuses, read logs, list runs, inspect versions, and manage running jobs. Use when the user asks about job status, wants to see logs, check if something is running, debug a failed job, or manage jobs on Datatailr / dt.

Roll back a Datatailr job (app, service, workflow, Excel add-in) to its previous version. Use when the user asks to roll back, revert, undo a deploy, go back to an earlier version, or restore the previous version of a job on Datatailr / dt.

Use Datatailr's user key-value store and secrets manager. KV stores user-owned configuration data. Secrets stores encrypted sensitive values like passwords, API keys, and tokens. Use when the user asks to store or retrieve user config, secrets, API keys, passwords, or values on Datatailr / dt. For platform admin settings (autoscaler, login, ACLs), use datatailr-admin instead.

Manage compute nodes on the Datatailr platform with the Node class of the Python SDK or the dt node CLI. List and inspect nodes, see the containers that run on a node, drain a node, read and set node tags, look up cloud instance types, and start, stop, restart, resize, create or terminate nodes. Use when the user asks about nodes, instances, machines, VMs, cluster capacity, draining, GPU nodes, or instance types on Datatailr / dt. For autoscaler and platform settings, use datatailr-admin instead.

Structure Python projects for Datatailr so shared code is importable and every job deploys without copying packages. Use when creating or restructuring a Datatailr project, adding a workflow/service/app/Excel add-in that shares code, fixing ImportError on deploy, or replacing sync/copy scripts for common modules.

Manage users and groups on the Datatailr platform. Create, list, inspect, and remove users and groups. Add or remove users from groups. Use when the user asks to manage platform users, create accounts, set up teams or groups, check who has access, or manage membership on Datatailr / dt.