feat(lightning-ai): dollar credit quotas and date parsing

Recover the Lightning AI provider from dangling git objects (lost during
linear rebase) and archive other retired fork-only providers before they
are garbage-collected.

Live provider:
- lightning_ai_provider.py: OpenAI-compatible API with dollar-based
  monthly credit quota tracking via /v1/memberships
- lightning_ai_quota_tracker.py: Balance fetching, caching, and
  cent-scale conversion for UsageManager integration

Archived to _retired/ (historical reference only):
- Cursor provider + quota tracker (sidecar architecture)
- Gemini A2A provider + client/session/sidecar/translator
- Bedrock provider stub
- Gemini file logger (superseded by transaction_logger)

Author: b3nw

.env.example

@@ -498,6 +498,27 @@
 #KILO_SESSION_TOKEN=""
 #KILO_QUOTA_REFRESH_INTERVAL=600
 
+# --- Lightning AI ---
+# Lightning AI uses an OpenAI-compatible API with dollar-based monthly credits.
+# API keys are UUIDs obtained from https://lightning.ai (Settings → API Keys).
+#
+# Key format: bare UUID (e.g. "a1b2c3d4-e5f6-7890-abcd-ef1234567890")
+# Optional billing routing suffix: "UUID/ORG_NAME/TEAMSPACE_NAME"
+#   - Bare keys charge against org-level credits (credits must be allocated at org level)
+#   - Keys with suffix route billing to that org/teamspace
+#
+# The proxy fetches credit balance from the /v1/memberships API using the
+# same API key — no separate session token is needed.
+#LIGHTNING_AI_API_KEY_1="your-lightning-ai-uuid"
+#LIGHTNING_AI_API_BASE="https://lightning.ai/api/v1"
+#
+# Monthly credit grant in whole dollars (used for quota bar display).
+# Plans: free=$15 (default), pro=$20, teams=$50
+#LIGHTNING_AI_MONTHLY_GRANT=15
+#
+# How often to refresh the credit balance in seconds (default: 300)
+#LIGHTNING_AI_QUOTA_REFRESH_INTERVAL=300
+
 # ------------------------------------------------------------------------------
 # | [ADVANCED] Debugging / Logging                                              |
 # ------------------------------------------------------------------------------

.fork/features/lightning-ai.md

@@ -34,3 +34,23 @@ Notes:
   compatibility. The TUI detects the `credits($)` group suffix and formats as dollars.
 - Retired providers are archived with retirement headers and documented in _retired/README.md.
 - All archived code was recovered before `git gc` could prune the dangling objects.
+
+## 2026-06-23 — Add Lightning AI documentation to .env.example and README.md
+
+Target: `feat(lightning-ai): dollar credit quotas and date parsing`
+Files:
+- `.env.example` (added Lightning AI section with env var reference)
+- `README.md` (added provider to Additional Providers table and quick-start env block)
+
+Verification:
+- llm-proxy-dev stack confirmed Lightning AI active with 48 models discovered
+- Health endpoint shows `lightning_ai` in active providers list
+- Provider routes requests to `https://lightning.ai/api/v1/chat/completions` and `/responses`
+
+Notes:
+- Resolves GitHub issue #70 (documentation gap for Lightning AI provider)
+- Documents: `LIGHTNING_AI_API_KEY_1` (UUID format), `LIGHTNING_AI_API_BASE`,
+  `LIGHTNING_AI_MONTHLY_GRANT` (plan tiers: free=$15, pro=$20, teams=$50),
+  `LIGHTNING_AI_QUOTA_REFRESH_INTERVAL` (default 300s)
+- Documents billing routing suffix (`UUID/ORG_NAME/TEAMSPACE_NAME`)
+- Notes that quota uses the API key directly (no separate session token like KiloCode)

README.md

@@ -28,6 +28,7 @@ A personal fork of [Mirrowel/LLM-API-Key-Proxy](https://github.com/Mirrowel/LLM-
 | **Command Code** | Bypasses standard subscription tier limits on chat completions by routing to the CLI endpoint (`/alpha/generate`). Supports dollar credits tracking mapped to cents baseline, 5-minute background refresh, and reasoning/thinking stream translation for `deepseek-v4-pro` and `mimo-v2.5-pro` |
 | **Umans** | OpenAI-compatible aggregation provider with static model definitions via `UMANS_MODELS`, request-based quota tracking (code_pro: 200 req/5h, max: unlimited), upstream-authoritative context windows, and cost tracking via sub-provider alias resolution |
 | **xAI Grok** | OAuth2 PKCE + Device Code flow for SuperGrok/X Premium+ accounts. Dual-endpoint model discovery from `api.x.ai` and `cli-chat-proxy.grok.com`, automatic token refresh, and CLI proxy context window metadata injection |
+| **Lightning AI** | OpenAI-compatible API with dollar-based monthly credit quota tracking via `/v1/memberships`. Dynamic model discovery, balance tracked in cents for integer quota precision. Plans: free ($15/mo), pro ($20/mo), teams ($50/mo) |
 
 ### Smart "Latest" Model Aliases
 
@@ -639,6 +640,12 @@ COMMAND_API_KEY_1=user_...  # Long-lived API key from CLI authentication
 KILO_SESSION_TOKEN=...  # Auto-refreshes on use, ~30-day TTL
 KILO_QUOTA_REFRESH_INTERVAL=600  # optional, default 600s
 
+# Lightning AI (dollar credit quota — balance fetched via API key, no separate token)
+# Key format: bare UUID or UUID/ORG_NAME/TEAMSPACE_NAME for billing routing
+LIGHTNING_AI_API_KEY_1=a1b2c3d4-e5f6-7890-abcd-ef1234567890
+LIGHTNING_AI_MONTHLY_GRANT=15  # optional, whole dollars: free=15, pro=20, teams=50
+LIGHTNING_AI_QUOTA_REFRESH_INTERVAL=300  # optional, default 300s
+
 # Umans (OpenAI-compatible aggregation provider)
 UMANS_API_KEY_1=your-umans-api-key
 UMANS_API_BASE=https://api.code.umans.ai  # optional, this is the default