v2.0.5: NVIDIA NIM fixes, statusline live-update, config routing updates
- Fix NVIDIA MODEL_MAP: stepfun/step-3-7-flash -> stepfun-ai/step-3.7-flash (live API catalog)
- Fix NVIDIA base_url: add /chat/completions (was causing 404)
- Add NVIDIA model slug_overrides to config.yaml and openrouter.py KNOWN_SLUGS
- Rewrite manual_statusline_start.py: live-updating single-line display (2s refresh, \r overwrite)
- Update test_resolve_route.py for current routing config
- Update config.yaml routing to user's current setup

feat: add NVIDIA NIM provider with 7 models and Nemotron thinking sup...

feat: add Sapiens AI (Agnes) provider with agnes-2.0-flash and agnes-...

fix: add quota fetch failure backoff (3 failures → 10min cooldown) an...

v2.0.1 - Major restructure: 8 new providers, per-provider context lim...

fix: resolve cmd.exe delims== parsing bug that nuked venv on every st...

feat: replace Flask dev server with Waitress for long-running stability

feat: increase body limit to 20MB, add 413 rejection, 429 backoff for...

feat: auto-detect and fix MiniMax MCP config on startup (#8)

chore: resolve .gitattributes merge conflict