Feat/cluster creator UI shortcut

[ikrispin]

Summary

Add UI widget detection to the cluster-creator skill and update CLAUDE.md with unified instructions that work across both UI-capable (ChatGPT with MCP Apps) and text-only (Claude Code, CLI) environments.

Changes:

1. cluster-creator SKILL.md: Added open_cluster_creator to allowed-tools frontmatter and replaced Step 1 with an environment detection step. When the open_cluster_creator tool is available, it is called immediately to open the interactive Creator widget form, skipping text-based parameter gathering (Steps 2-6). Text-only environments fall back to the existing step-by-step workflow unchanged.

2. CLAUDE.md: Replaced the previous multi-persona instructions with a focused OCP Admin instructions file that includes a UI Widget Policy section. The policy directs the agent to call UI-opening tools (open_cluster_creator, list_clusters, get_cluster_hosts) immediately when interactive widgets are available, while gracefully falling back to skill-driven text workflows in text-only environments. This file serves as the unified agent instructions across ChatGPT (custom GPT instructions), Claude Code (CLAUDE.md), and other agent platforms.

Pack(s) affected

• ocp-admin
• rh-ai-engineer
• rh-automation
• rh-basic
• rh-developer
• rh-sre
• rh-virt
• Other / repo-wide

Change type

• New skill
• New agent
• New pack
• Update existing skill / agent
• MCP server config (mcps.json)
• Docs / README
• CI / tooling
• Federation (external pack)

Contribution method

• Created/imported with /agentic-contribution-skill
• Manual contribution (validated with make validate + make validate-skill-design-changed)

Pack-persona alignment (new skills only)

N/A — updating existing skill and agent instructions.

CLAUDE.md compliance

• Agents orchestrate skills; no direct MCP/tool calls in agents
• Skills are single-purpose task executors
• Skills encapsulate all tool access (MCP tools invoked only inside skills)
• Document consultation: file is read with the Read tool, then declared to the user
• No credentials hardcoded; env vars used via ${VAR} references
• Human-in-the-loop confirmation added for any destructive or critical operations

Validation

• make validate passes locally
• New/changed skills have valid YAML frontmatter (name, description)
• New/changed agents have valid YAML frontmatter (name, description)

[dmartinol]

I'm fine with the fix but UI tools will never be available until we release them and publish the assisted-service-mcp OCI image (that's the reason for the CI pipeline failure, BTW).
Do you prefer to keep this on hold until we agree on a release strategy for the MCP or do you need to have it available right now?

[ikrispin]

I'm fine with the fix but UI tools will never be available until we release them and publish the assisted-service-mcp OCI image (that's the reason for the CI pipeline failure, BTW). Do you prefer to keep this on hold until we agree on a release strategy for the MCP or do you need to have it available right now?

exactly right, I'll make it a draft for now.

[dmartinol]

Thank you for your contribution!
This repo will be soon archived and we're not accepting new commits.
Please move these changes to the new location at https://github.com/RHEcosystemAppEng/agentic-collections-skills