Skip to main content
A workflow is a named, reusable command snippet stored in Warp Drive. Workflows let you save commands you run repeatedly — with optional parameters, descriptions, and tags — and run them again from any Warp session without retyping. You can keep workflows private, share them with your team, or use AI-generated workflows suggested in response to natural-language queries.

​
Workflow types

Warp supports two kinds of workflows:

Command workflows

A shell command with optional parameterized arguments. When you run a command workflow, Warp inserts the command text into the input area — with any arguments pre-filled or prompted.

Agent Mode workflows

A natural-language query that is sent to Warp’s AI agent rather than run directly as a shell command. Useful for repeatable AI tasks like “summarize the git log since last release.”

​
Creating a workflow

1

Open the workflow creator

You can create a workflow in several ways:
  • Click Save as Workflow in any block’s toolbar to pre-populate the creator with that block’s command.
  • Open the command palette (Cmd+P / Ctrl+P) and search for Create workflow.
  • Open the workflow library (via the left sidebar or Cmd+Shift+R) and click New workflow.
2

Fill in the details

Give your workflow a name — this is how you’ll find it later. Optionally add:
  • Description — a plain-English explanation of what the workflow does.
  • Tags — to organize workflows by category (e.g., git, docker, deployment).
  • Shell restrictions — limit the workflow to specific shells (bash, zsh, fish, PowerShell) if it uses shell-specific syntax.
3

Add arguments (optional)

Replace variable parts of the command with argument placeholders using the {{argument_name}} syntax. For each argument, you can set:
  • A name and description shown in the fill-in UI.
  • An argument type: free text, a file path, or an enum (a predefined list of values from Warp Drive).
  • A default value that is pre-filled when the workflow runs.
4

Save and sync

Click Save. Local workflows are saved immediately. Cloud workflows (personal or team) are synced to Warp Drive and available on all your devices.

​
Parameterized workflows

Parameterized workflows let you define a command template where certain values are filled in at runtime. For example: 
git checkout -b {{branch_name}}
When you run this workflow, Warp prompts you to enter a value for branch_name before inserting the command. You can also define enum-type arguments to show a dropdown of allowed values instead of a free-text input. 
kubectl get pods -n {{namespace}} --context {{cluster}}
Enum arguments pull their list of values from a Warp Drive enum object. This means all teammates using the same workflow see the same up-to-date list of allowed values — useful for namespaces, environments, or region codes that change over time.

​
Running a workflow

You can run any saved workflow from several entry points:
  • Command palette — open with Cmd+P / Ctrl+P, type the workflow name or a keyword, and press Enter.
  • Workflow library — open with Cmd+Shift+R / Ctrl+Shift+R. Browse by category, search, or filter by source.
  • Slash menu — type / in the input area to open a quick-access menu of recent workflows.
  • Up-arrow history — if a workflow was previously run, it appears in up-arrow history alongside regular commands.
  • Agent Mode — Warp AI can suggest and insert relevant workflows as part of an agent session.

​
Workflow sources

Workflows can come from several sources:
SourceDescription
localStored in a local YAML file on your machine
projectStored in a .warp/workflows/ directory in a git repo, shared via version control
personal_cloudSynced to your Warp account, available on all your devices
teamShared with a Warp team (requires a team plan)
globalCurated community workflows shipped with Warp
warp_aiEphemeral workflows generated by Warp AI — not persisted unless you save them
notebookWorkflows embedded inside a Warp notebook document

​
Workflow file format

Local and project workflows are stored as YAML files. A command workflow looks like this: 
name: "Deploy to staging"
command: "kubectl apply -f {{manifest_path}} --context staging"
description: "Apply a Kubernetes manifest to the staging cluster"
tags:
  - kubernetes
  - deployment
arguments:
  - name: manifest_path
    description: "Path to the YAML manifest file"
    default_value: "./k8s/deployment.yaml"
An Agent Mode workflow looks like this: 
type: agent_mode
name: "Summarize recent git activity"
query: "Summarize the git commits from the last {{days}} days and list any breaking changes"
arguments:
  - name: days
    description: "Number of days to look back"
    default_value: "7"

​
Sharing workflows

To share a workflow with your team:
  1. Open the workflow in the workflow library.
  2. Click Share and choose a team destination, or copy the workflow as YAML to paste into a pull request.
  3. Team members with access to the same Warp team will see the shared workflow appear automatically in their workflow library.
Workflows shared at the project source level live alongside your code in version control. Add a .warp/workflows/ directory to your repository and commit YAML workflow files there — any Warp user who opens that repository will automatically see those workflows.

​
Workflow aliases

Warp also supports shell aliases as a lightweight alternative to workflows. Aliases defined in your shell’s rc file (.bashrc, .zshrc, etc.) are resolved by the completion engine and appear in history. However, aliases cannot be parameterized, shared via Warp Drive, or run from the command palette — for those use cases, use a proper workflow.

​
The workflow library

The workflow library is the central browsing UI for all your workflows. It is organized into categories (from categories.rs in the workflows module) that group workflows by topic or source. Use the search box at the top to filter by name, description, or tag.
You can export a workflow to YAML from the workflow library by clicking the Export button on any workflow that supports export. This is useful for migrating workflows into version control or sharing them outside of Warp.

Build docs developers (and LLMs) love

Get started for freeTalk to us