Production-ready AI skills and automation for Red Hat platforms — specialized plugins for SREs, developers, platform administrators, and AI engineers working with RHEL, OpenShift, and Red Hat automation platforms.
Browse the catalog: The published marketplace UI is at rhecosystemappeng.github.io/agentic-catalog — powered by agentic-catalog. This repo is where skills are authored and maintained. The catalog aggregates content from here periodically.
| Pack | Skills | Description | Persona |
|---|---|---|---|
| rh-sre | 13 | CVE remediation, system compliance, RHEL automation | Site Reliability Engineers |
| rh-developer | 14 | Application deployment, S2I builds, Helm charts | Application Developers |
| rh-virt | 10 | VM lifecycle, snapshots, migrations, cloning | Virtualization Admins |
| ocp-admin | 3 | Multi-cluster management, health reports, monitoring | OpenShift Administrators |
| rh-ai-engineer | 11 | Model serving, vLLM, KServe, NVIDIA NIM | AI/ML Engineers |
| rh-automation | 11 | Ansible Automation Platform governance, safety checks | Automation Leads |
| rh-basic | 6 | CVE diagnostics, lifecycle, patching, support cases | General Red Hat Users |
See each pack's README for available skills and usage examples.
The eval/ directory contains skill evaluation reports (report.json + report.md per skill) generated by an external evaluation pipeline.
Pre-packaged ZIP files are available on the Releases page. The latest release is auto-updated on every merge to main. Each ZIP contains all skills for a pack with documentation included.
To use them:
- Go to the latest release
- Download the ZIP for your pack (e.g.,
ocp-admin.zip) - Upload to your AI agent of choice (e.g., ChatGPT Custom GPT, Claude)
- Claude Code (latest version recommended) or Cursor
- Container Runtime (Podman or Docker) for MCP servers
- Network Access to Red Hat platforms (depending on packs used)
- Credentials: Red Hat API keys, cluster kubeconfigs, service tokens (per pack)
New skill idea? Run /agentic-contribution-skill in Claude Code.
Have an existing skill? Run /agentic-contribution-skill and choose import mode.
Full guide:CONTRIBUTING.md
# Install dependencies (first time)
make install
# Validate pack structure + compliance
make validate
# Validate only changed skills (recommended for local dev)
make validate-skill-design-changed
# Validate all skills or a specific pack
make validate-skill-design
make validate-skill-design PACK=rh-sre
# Validate MCP tool references against live servers
make validate-mcp-toolsSkills are validated against Design Principles:
- Document consultation transparency (DP1)
- Parameter specification and ordering (DP2)
- Description conciseness (DP3)
- Dependencies declaration (DP4)
- Human-in-the-loop requirements for critical operations (DP5)
- Mandatory sections (Prerequisites, When to Use, Workflow) (DP6)
- Credential security (no
echo $VARexposure) (DP7) - Required frontmatter policy (
modelrequired:inherit|sonnet|haiku;colorrequired:cyan|green|blue|yellow|red|magenta)
This repository uses gitleaks and pre-commit to block accidental secrets.
# One-time setup: install deps + git hooks
make install
scripts/install-hooks.shMCP Configuration Rules:
{
"env": {
"LIGHTSPEED_CLIENT_ID": "${LIGHTSPEED_CLIENT_ID}",
"LIGHTSPEED_CLIENT_SECRET": "${LIGHTSPEED_CLIENT_SECRET}"
}
}Never hardcode credentials. Always use ${ENV_VAR} references. See SECURITY.md for details.
- CLAUDE.md: Repository structure and development workflow
- Skill Design Principles: Quality guidelines for skills
- Security Policy: Credential handling and vulnerability reporting
Apache License 2.0 - see LICENSE for details.
Maintained by:Red Hat Ecosystem Engineering