CrewAI Skills - Setup Guide¶

Source: crewAIInc/skills (44⭐, no LICENSE file; last push Jun 16, 2026) Skill family: crewAIInc/skills (4 SKILL.md files in-repo; 4 indexed listings) Publisher: CrewAI (crewAIInc) - the team behind the CrewAI framework (59.3K⭐) Category: AI Agent Frameworks / Agent Design Quality Tier: 🟡 Beta (official publisher; low repo stars, no LICENSE file, and no upstream pushes since Jun 16, 2026 - verify terms before production reliance; verified Oct 4, 2026)

Four official skills that teach AI coding agents how to build with CrewAI, the role-playing multi-agent framework. The set covers architecture decisions and project scaffolding (getting-started), agent configuration (design-agent), task design (design-task), and a live documentation lookup (ask-docs). Each skill ships extra reference files for progressive disclosure, and the repo registers a Claude Code plugin marketplace alongside the standard Agent Skills layout used by Hermes.

Installation¶

# Full suite
npx skills add crewAIInc/skills

# Or install a single skill
npx skills add crewAIInc/skills --skill design-agent

Claude Code also supports the repo's built-in plugin marketplace:

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

Roster - All 4 Skills¶

Skill	Installs	Does
getting-started	7,521	Architecture decisions + scaffolding: when to use LLM.call / Agent.kickoff / Crew.kickoff / Flow, `crewai create flow`, YAML config (agents.yaml, tasks.yaml), @CrewBase wiring, @start / @listen Flows, conversational Flows, variable interpolation
design-agent	7,469	Agent design: Role-Goal-Backstory framework, LLM selection, tool assignment, execution tuning (max_iter, max_rpm, max_execution_time), memory + knowledge sources, guardrails, YAML vs code config
ask-docs	7,350	Documentation lookup: queries the official docs at docs.crewai.com for API details, advanced features, troubleshooting, enterprise features, and tool references beyond the other three skills
design-task	7,326	Task design: descriptions + expected_output, context dependencies, structured output (output_pydantic / output_json / output_file), guardrails, human-in-the-loop review, async execution

Reference files: getting-started ships conversational-flows / flow-routing / mcp-servers / tools-catalog; design-agent ships custom-tools + memory-and-knowledge; design-task ships structured-output.

CorpusIQ Use Cases¶

Use Case	How
Agent architecture reference	The getting-started decision tree (call vs kickoff vs flow) is a compact framework for scoping multi-agent builds, including CorpusIQ's own automations
Skill authoring pattern	The family is a clean reference implementation of progressive disclosure (main SKILL.md + `references/`) and multi-host packaging (`.claude-plugin` marketplace + portable Agent Skills layout)
Agent config checklists	design-agent / design-task work as ready-made review checklists for any CrewAI-style setup: role clarity, tool scoping, output schemas, guardrails

Integration with CorpusIQ¶

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects

Limitations / Verification¶

Verified Oct 4, 2026: 4 indexed listings on skills.sh; 4 SKILL.md files confirmed via the GitHub trees API (branch main); ~29,666 combined installs.
No LICENSE file in the repo (GitHub license API returns 404, checked Oct 4, 2026) - confirm terms before commercial use.
The skills repo is small (44⭐, 20 forks) next to the main CrewAI framework repo (59.3K⭐); last upstream push Jun 16, 2026.
The skills teach CrewAI project patterns rather than running standalone tasks; ask-docs needs access to docs.crewai.com.
No live install test performed; install counts are from the Oct 4, 2026 sweep snapshot.

Security¶

skills.sh security verdicts (verified Oct 4, 2026) - all four skills pass all three checks:

Audit	Verdict
Gen Agent Trust Hub	Pass
Socket	Pass
Snyk	Pass

← Skills Catalog | Skills Marketplace → Powered by CorpusIQ