Update lab web link using builder project id

Make site count telemetry authoritative and tested

Improve site count parsing and lab telemetry

Align site naming and auto-config with dynamic store builds

Support configurable site counts in GNS3 retail builder

Force GNS3 build to honor site count and retry if missing

Remove diagram UI and enforce requested site count/GNS3 build

Hardening: pin Anthropic by default, add fallbacks and cached demo

Try Blaxel agents/query endpoint first

Fallback to Anthropic/OpenAI when Blaxel fails

Add Blaxel endpoint fallbacks

Add judge quickstart and update Blaxel headers

Default to Blaxel (env override OG_LLM_PROVIDER)

Improve WG bring-up diagnostics and persist mermaid debug

Fix deployment call signature and simplify mermaid labels

Sanitize mermaid labels and log diagram text for debugging

Sanitize mermaid labels to avoid syntax errors

Render mermaid diagrams without emojis/HTML for compatibility

Use LLM-assisted human-readable project names with fallback

Run SSH even after seeding in hybrid mode; allow seeding toggle

Add WG bring-up status, runtime log tail, and mermaid image rendering

Bring up WireGuard when provided and add SSH preflight summary

Skip SSH when seeding succeeds in hybrid mode; add SSH-on-seed-fail flag

Fix hybrid seeding to use template engine and add generate_config alias

Add hybrid GNS3 seeding for deployment and enable deploy by default

Re-enable deployment by default and attach lab info to runs

Disable real deployment by default; require env flag to enable

Create GNS3 projects per run and surface dynamic lab links

Shorten MCP call timeout and make get_projects non-blocking

Add brownfield import mode and sponsor LLM support

Normalize device roles/vendors and render mermaid diagram

Fix syntax error in Hugging Face response handling

Fix f-string escapes for new provider endpoints

Remove OpenRouter placeholder code

Add Blaxel, SambaNova, Nebius, and Hugging Face LLM providers

Disable OpenRouter; only Anthropic/OpenAI providers remain

Normalize LLM env vars to strip whitespace and report presence accurately

Remove LLM mock fallbacks and surface real errors

Log full exception details for Anthropic errors

Surface API errors and fallback to OpenAI if Anthropic fails

Improve consultation flow, diagrams, and fallback validation readiness

Improve monitoring UX and deterministic offline outputs

Fix ray module import error and add interactive consultation UI

Fix f-string syntax error in suzieq_client.py

Add comprehensive API monitoring and cost tracking

test: Stage 6 comprehensive test suite - 23/23 passing

feat: Stage 6 - Autonomous Deployment Engine

feat: Ray distributed execution for hyperscale deployments

feat: GitOps workflow engine with environment promotion (Todo #6)

feat: RAG-based incident learning system (Todo #5)