PlugWiki

getting-started skill (crewAIInc): scaffold and wire a CrewAI project

crewAI's own skill that guides a coding agent through choosing the right CrewAI building block and scaffolding a project with the crewai CLI.

Installs
8.6K
on skills.sh
Stars
48
on GitHub
Files
5
besides SKILL.md
Updated
Oct 2026
last repo commit

getting-started makes your coding agent follow CrewAI's own scaffolding rules: run crewai create flow first, then edit the generated files, instead of writing the project by hand. It also teaches how to pick between a single LLM call, one agent, a crew or a Flow, and how to wire YAML agents and tasks to a Python crew class.

It loads when you start a CrewAI project, set up agents.yaml and tasks.yaml, or build a chat-style Flow. It needs the crewai CLI, and running the result calls your model provider with your own API key.

Good for

  • Python developers building agents or crews with the CrewAI framework
  • People new to CrewAI who want a first project set up the supported way
  • Developers choosing between Agent.kickoff(), Crew.kickoff() and a Flow

Install getting-started

Claude Code

  1. 1

    Copy this prompt.

    Install the getting-started skill in Claude Code. Run this command:

/plugin marketplace add crewAIInc/skills
/plugin install crewai-skills@crewai-plugins

Then check that it loaded and tell me how to use it.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Pick one route, not both, or you get every skill twice. The README documents the plugin route for Claude Code. The plugin installs all 10 skills in the crewai-skills plugin; the skills CLI route installs getting-started, ask-docs, build-flow. getting-started tells the agent to re-check version-sensitive details with ask-docs and sends Flow work to build-flow. The other CrewAI skills are optional hand-offs.

From the project README. Source.

What's in the skill

When it loads

CrewAI architecture decisions and project scaffolding. Use when starting a new crewAI project, choosing between LLM.call() vs Agent.kickoff() vs Crew.kickoff() vs Flow, scaffolding with 'crewai create flow', setting up YAML config (agents.yaml, tasks.yaml), wiring @CrewBase crew.py, writing Flow main.py with @start/@listen, building experimental conversational Flows with handle_turn()/chat(), or using {variable} interpolation.

Sections in SKILL.md

  1. MANDATORY WORKFLOW — Read This First
  2. 1. Choosing the Right Abstraction
  3. Decision Flowchart
  4. 2. LLM.call() — Direct LLM Invocation
  5. 3. Agent.kickoff() — Single Agent Execution
  6. 4. CLI Scaffold Reference
  7. 5. Crews: YAML Config and @CrewBase
  8. 6. Flows - The Production Foundation
  9. Basic Flow — main.py
  10. Using Agent.kickoff() Inside Flows (Common Pattern)
  11. Mixing Abstractions in a Flow
  12. Routing, Joins, Persistence, Human Feedback - the Short Version
  13. Conversational Flows with handle_turn()
  14. 7. Variable Interpolation with inputs
  15. 8. Running Your Project
  16. 9. Quick Diagnostic Checklist

Key facts

Author
crewAI (the crewAIInc organisation); the plugin manifest names crewAI as author and the repo is the vendor's own
License
MIT (LICENSE file and README agree)
Installs on skills.sh
8,575 as of 2026-11-02
GitHub stars
48 for the whole crewAIInc/skills repo, 2026-11-02
What is inside
6 files: SKILL.md and 5 files in references/ (crew YAML config, flow routing, conversational flows, MCP servers, tools catalog). It bundles no scripts.
Last change to this skill
2026-10-06
Part of
The crewai-skills plugin (10 skills) in the crewai-plugins marketplace
Checked against
crewai 1.15.23 on 2026-10-01, per the SKILL.md

What it includes

  • Run crewai create flow, edit the files, then crewai install and crewai run

    SKILL.md says never to write main.py, crew.py, agents.yaml, tasks.yaml or pyproject.toml by hand. The agent runs the CLI scaffold first and then edits what it generated.

    Try: “Set up my first CrewAI project from scratch.”

    Runs code · Mandatory scaffold workflow

  • Five levels from LLM.call() to a conversational Flow

    A table and flowchart pick between LLM.call(), Agent.kickoff(), Crew.kickoff(), a Flow around crews, and a conversational Flow. For production apps it says to start with a Flow.

    Read · Choose the abstraction

  • Wire agents.yaml and tasks.yaml to a crew class

    Covers {variable} placeholders, matching method names to YAML keys, and sequential versus context-based task input.

    Write · Crew YAML and @CrewBase

  • @start, @listen and @router steps with shared state

    Short guide to Flow state, routing, or_ and and_, @persist and @human_feedback, with the full API left to the build-flow skill.

    Write · Flows

  • Chat sessions where each message re-runs a Flow

    Uses handle_turn(message, session_id=...) per user line, or flow.chat() for a terminal. SKILL.md labels this experimental.

    Write · Conversational Flows

  • Pass {topic}-style inputs through kickoff()

    Explains how flow inputs reach state, crew inputs and YAML, and the common mistakes such as missing inputs or Jinja braces.

    Read · Variable interpolation

  • A symptom, cause and fix table for common setup errors

    Covers errors like a missing template variable, a method name that does not match a YAML key, and missing provider extras.

    Read · Diagnostic checklist

  • Five deeper guides the agent can open

    Crew YAML configuration, flow routing and persistence, conversational flows, MCP servers and a tools catalog.

    Read · references/

From the official docs

Permissions

Medium risk

The agent runs the crewai CLI, creates a project folder and runs your code, which calls your model provider with your own API key. Check the generated .env and the provider bill before you run the result.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • Running the project spends your model provider credits

    With no MODEL set, agents use gpt-4.1-mini and need OPENAI_API_KEY. Set a spending limit with your provider before running agent loops. github.com

  • It tells the agent to always scaffold with the CLI

    The agent will run crewai create flow rather than write files by hand, so run it in an empty folder. github.com

  • The plugin route installs all 10 skills

    The crewai-skills plugin manifest lists 10 skills and holds no hooks, agents or MCP config in the git tree. Use the skills CLI route if you want only the skills this page names. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details
Talks toYour model provider (set by MODEL and its API key). Package indexes when crewai install or uv add runs
Sign-in and keysThe scaffold's .env holds an OPENAI_API_KEY placeholder that you replace; SKILL.md says never to commit it.
Data retentionNot stated in SKILL.md.
Scopes

  • Runs crewai create flow, crewai install and crewai run, and uv run kickoff

  • Creates a project folder with src/, config YAML, a .env placeholder and AGENTS.md, CLAUDE.md, GEMINI.md and CURSOR.md files

  • SKILL.md mentions uv add for provider extras such as crewai[anthropic]

Questions

What does the getting-started skill do?

It guides your coding agent through starting a CrewAI project: choosing between LLM.call(), Agent.kickoff(), Crew.kickoff() and Flows, scaffolding with the crewai CLI, and wiring YAML config to a crew class.

How do I install getting-started?

Use the install box at the top of this page. The README documents a Claude Code plugin that installs all 10 CrewAI skills. The skills CLI route in the box installs getting-started plus ask-docs and build-flow, which SKILL.md names in its body.

Does it need the other CrewAI skills?

Only partly. SKILL.md names ask-docs for re-checking version-sensitive details and build-flow for the full Flow API, so the skills CLI route installs those two. Its related-skills list also names check-crewai-api, connect-tools-and-mcp, design-agent, design-task, test-crewai-project, deploy-to-amp and call-deployed-crew; add them when you need that depth.

Does it need an API key?

The skill itself needs none, but the project it builds does. SKILL.md says agents default to gpt-4.1-mini and need OPENAI_API_KEY, or the key for the provider you set with MODEL.

Is getting-started safe to use?

It runs the crewai CLI and creates files in your project folder, so use a new or version-controlled folder. Running the scaffolded project then calls your model provider with your key.

Is this the official CrewAI skill?

Yes, it comes from the crewAIInc organisation's own skills repo, and the marketplace file describes the plugin as official crewAI plugins for Claude Code.

Does it work in the Claude app, Codex or Cursor?

The README documents Claude Code only. The skill runs the crewai CLI and writes project files, so there is no Claude app upload option here. The box lists the Codex skills folder as a generic route.

Information

updated Nov 2, 2026

Repository
crewaiinc/skills
Folder
skills/getting-started

crewaiinc/skillsskills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4
  1. README.md · github.com
  2. marketplace.json · github.com
  3. plugin.json · github.com
  4. SKILL.md · github.com

Updated Nov 2, 2026.