Added allowed-tools declaration to all 71 skills across 7 packs

[r2dedios]

Summary

Adds the allowed-tools field to the YAML frontmatter of all 71 skills across the 7 packs (ocp-admin, rh-ai-engineer, rh-automation, rh-basic, rh-developer, rh-sre, rh-virt). This field declares upfront which MCP tools each skill is permitted to use, giving the LLM an explicit boundary before it starts executing.

Skill declarations:

• 67 new declarations added to skills that had none
• 4 existing declarations updated for consistency (debug-rbac, debug-scc, incident-triage, vm-snapshot-create)
• Skills that use no MCP tools declare allowed-tools: with an empty value
• Orchestration skills list the transitive union of tools from all delegated sub-skills
• Uses short tool names (e.g. resources_get) for straightforward validation against live MCP servers
• Lightspeed-mcp tools use namespace prefixes (e.g. vulnerability__get_cves, inventory__find_host_by_name)

CI validator improvements (scripts/validate_mcp_tools.py):

• Tools from non-startable MCP servers (non-container commands like npx, uvx, or empty) are now classified as WARN instead of FAIL — they can't be verified, but they don't block the PR
• Added per-tool logging for easier debugging of tool name discrepancies
• Removed dead queried_servers variable

CI workflow changes (.github/workflows/mcp-tool-validation.yml):

• Added second dummy kubeconfig context so kubernetes-mcp-server exposes the configuration_contexts_list tool (requires 2+ contexts)

Documentation:

• Added mcp-tool-validation.yml section to .github/workflows/README.md (was missing)
• Updated workflow count from 5 to 6

Pack(s) affected

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

Change type

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

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
• All 3 CI checks pass: mcp-tool-check, compliance-check, skill-linter
• New/changed skills have valid YAML frontmatter (name, description)
• allowed-tools format validated (space-delimited, short names)

[r2dedios]

@dmartinol PR ready for review!

[dmartinol]

Thank you! /lgtm