Add multi-agent collaborative mode: Investigator + Responder roles with /ws/multi endpoint

Add curriculum learning: adaptive task difficulty scheduling via /curriculum endpoints

docs: update Benchmark Results with real baselines.py scores

Fix /leaderboard endpoint: add agent_comparison benchmark data

Add baseline agent comparison: Random, RuleBased, InformationFirst vs GRPO-LLaMA

docs: update README with real benchmark scores from baselines.py

Fix /tasks endpoint in server/app.py: add dns, waf, thundering_herd; update /about task count to 10

Fix baselines.py scoring bug: use grade_episode(), fix seed variation; fix api.py missing tasks

Add baselines.py: three baseline agent comparison; fix runbook UTF-8 encoding

Add Tasks 8-10: Real incident post-mortems (GitHub DNS, Cloudflare WAF, Slack Thundering Herd)

final: submission cleanup — remove junk files, update README endpoints, clean .gitignore

feat: rewrite heuristic agent with task-specific optimal sequences — all 7 tasks now score >0.75 and resolve

fix: /live dashboard JS — state field names, memory field, alert severity

fix: auto-load .env for HF_TOKEN + fix CPU inference device bug

feat: replay system, human challenge, progress viz, 2 new generator failure modes

fix: switch NOC dashboard from WebSocket to REST polling to sync with live demo_llm.py execution

feat: /live WebSocket NOC dashboard — cinematic real-time incident visualization

fix: load LoRA adapter via PEFT instead of merged 16GB model

fix: correctly handle PEFT adapter and token parsing for transformers fallback

fix: add transformers fallback for unsloth module in demo_llm.py

fix: rewrite frontmatter, single-line short_description

fix: shorten short_description under 60 chars

fix: remove kaggle_training checkpoints from git, add to gitignore

Finalizing ARIA for production deployment: 8B model migration, documentation polish, cleanup

feat: /about endpoint, stunning dashboard UI, auto-loop demo, rich /validate

feat: stunning ARIA dashboard — complete UI overhaul

feat: add demo_llm.py — heuristic + LLM agent Rich TUI demo

fix: shorten short_description to meet HF 60-char limit

Merge branch 'main' of https://github.com/Twilight-13/devops-incident-response

chore: remove model checkpoints from Space repo (use LFS or model repo instead)

feat: final submission — README frontmatter, BLOG, training notebook, all ARIA features

feat: final submission — README frontmatter, BLOG, training notebook, all ARIA features

Upload README.md with huggingface_hub

Upload openenv.yaml with huggingface_hub

Upload server/app.py with huggingface_hub

feat: optimize for LLM evaluation — README, openenv.yaml, /about endpoint, endpoint docstrings

fix: sync all endpoints to production entrypoint, clean junk files

fix: add multi_agent package to HF Space

fix: add curriculum + multi-agent routes, fix VALID_TASKS exclude generated, fix title

fix: fix validate exclude generated, fix dashboard title

feat: ARIA Unified Dashboard, Blog, and Project Identity Overhaul

feat: synchronize Incident Generator to HF Space (server/app.py)

feat: ARIA Incident Generator — procedural incidents from seeds

Sync: all 7 tasks, openenv-core migration, training notebook, cleanup

Complete README overhaul: 7 tasks, training notebook, WebSocket, metrics documented

Add /metrics and /leaderboard endpoints for live episode tracking

Add GRPO training notebook demonstrating agent learning from environment

Add Task 7: Multi-Region Failover (partial failover with compliance constraints)

Add Task 6: Database Performance Degradation (missing index investigation)

Add Task 5: Security Incident Response (DDoS detection and mitigation)