[docs] Consolidate agent instructions: move universal content to AGENTS.md, pointer in CLAUDE.md

[jjshanks]

Contribution Type

• Recipe (/recipes)
• Schema (/schemas)
• Dashboard (/dashboards)
• Integration (/integrations)
• Skill (/skills)
• Repo improvement (docs, CI, templates)

What does this do?

Consolidates all agent instructions into AGENTS.md as the single source of truth. CLAUDE.md is now a thin pointer that references AGENTS.md, with only the Claude Desktop-specific MCP connector path preserved. Removes duplicated content (repo structure, guard rails, PR standards, key files, GSD layer) between the two files.

Requirements

None — this is a docs-only change.

Checklist

• I've read CONTRIBUTING.md
• No credentials, API keys, or secrets are included

[alanshurafa]

Thanks for the contribution. Consolidating the universal repo-orientation content into AGENTS.md and leaving CLAUDE.md as a pointer is a sensible refactor — AGENTS.md is the tool-agnostic location, and it de-duplicates guidance that applies to every AI tool, not just Claude.

The change is a roughly balanced move (+60/-56). One thing for the maintainer to confirm, since this is the contributor-contract documentation: that the move is complete (no guidance dropped between the two files) and that the CLAUDE.md pointer is enough for Claude-specific tooling that may read CLAUDE.md first. Recommend maintainer review.

— Alan (community reviewer; non-binding)