[Enhancement]: Integrations layout — discoverable stacks (Ollama, APIs, frameworks, Skillware, …)

[rosspeili]

Problem or use case

Core AURA patterns and stack-specific demos are mixed together. Users and agents need a clear index to discover their stack (Ollama, OpenAI, Claude, Gemini, LangGraph, Skillware, plain Python, etc.) and see how AURA attaches.

Proposed behavior

Add top-level integrations/ — parallel to examples/:

integrations/
  README.md              # index: pick your stack → path → run command
  ollama/
    README.md
    llama_loop.py        # llama3.2:1b default; pattern works for any Ollama tag
  skillware/
    README.md
    mock_tools.py
    live_tools.py        # [skillware] extra
  openai/
    README.md
    chat_loop.py
  anthropic/
  google/
  langgraph/             # stub or link to future wrap issue

Rules:

• examples/ = core AURA patterns (minimal loop, guarded tools, task mode, sequencer)
• integrations/ = stack-specific demos, docs, optional notebooks — one folder per stack, equal treatment in the index
• Each integration README: prerequisites, .env vars, run command, what AURA records, customization (model swap, rules, coat posture)
• Link from docs/integrations/README.md, docs/INDEX.md, onboarding (#13)

Coordinate with Examples flat restructure draft — stack-specific content moves to integrations/<stack>/.

Alternatives considered (optional)

Keep everything under examples/ with prefixes — rejected; blurs core vs stack-specific.

[evgenyponomarev]

I'd like to work on this. Starting now - happy to hand it over if someone else is already on it or if maintainers would rather discuss the approach first.

(Sent by Sloppy on my behalf)

[rosspeili]

Hey @evgenyponomarev assigned <3

Some tips:

Already on main (don’t re-land): integrations/{skillware,openai,anthropic,google}/, cloud body loops, ollama_skill_loop.py, and the discoverability index at docs/integrations/README.md (see CHANGELOG #19 entry). That’s the bulk of the layout.

What #19 still needs (your PR #51 direction):

1. Top-level integrations/README.md — pick-your-stack index with path + run command (issue sketch).
2. integrations/ollama/ — split Ollama into its own stack folder (llama_loop.py + README); keep skillware README pointing at it.
3. Skillware entry scripts — mock_tools.py / live_tools.py (or equivalent thin wrappers) so the index matches the issue tree.
4. integrations/langgraph/ stub README (planned link is enough).
5. Doc ripple — docs/INDEX.md, getting-started.md, docs/integrations/README.md: core vs stack boundary (examples/ = patterns 01–04 flat after [Enhancement]: Restructure examples — flat named scripts and single README index #21; integrations/ = stack demos). Drop stale examples/04-sequencer-pipeline/ links once Closes #21: [Enhancement]: Restructure examples — flat named scripts and single README index #30 lands.

Coordination: please rebase onto current main (post-#45) and merge #30 (#21) first or at least avoid conflicting example-path edits. Examples 05–08 can stay under examples/ for this PR if docs clearly index them as integration/ToolHost demos; full physical move is optional follow-up.

Ping here when #51 is rebased, willll review against the checklist above.