chore(other): deno fmt

Author: algoflows

.vscode/settings.json

@@ -1,3 +1,3 @@
 {
   "claudeCodeChat.permissions.yoloMode": true
-}
+}

apps/classify/workflow/BATCHING.md

@@ -23,11 +23,13 @@ INTERNAL_EMIT_DELAY_MS=100
 ### Script Defaults
 
 All scripts use these defaults:
+
 - **Batch size:** 5 indicators per batch
 - **Concurrent batches:** 4 batches running in parallel
 - **Total concurrency:** 20 indicators processing simultaneously
 
 This is configured in:
+
 - [scripts/run-random.ts](./scripts/run-random.ts) (lines 418-419)
 - [scripts/run-all.ts](./scripts/run-all.ts) (lines 417-418)
 
@@ -61,6 +63,7 @@ This is configured in:
 ## Usage Examples
 
 ### Run Random Indicators
+
 ```bash
 # Process 20 random indicators (4 batches × 5)
 deno task run:random -- -20 openai
@@ -70,6 +73,7 @@ deno task run:random -- -100 openai
 ```
 
 ### Run All Indicators
+
 ```bash
 # Process all indicators (in groups of 20)
 deno task run:all openai
@@ -81,12 +85,13 @@ deno task run:all 40 openai
 ## Performance Tuning
 
 ### Increase Concurrency
+
 To process **40 indicators concurrently** (8 batches of 5):
 
 1. Update scripts:
    ```typescript
    const batchSize = 5;
-   const concurrentBatches = 8;  // Changed from 4
+   const concurrentBatches = 8; // Changed from 4
    ```
 
 2. Ensure your system can handle it:
@@ -95,15 +100,17 @@ To process **40 indicators concurrently** (8 batches of 5):
    - Database connection pool may need adjustment
 
 ### Reduce Concurrency
+
 To process **10 indicators concurrently** (2 batches of 5):
 
 1. Update scripts:
    ```typescript
    const batchSize = 5;
-   const concurrentBatches = 2;  // Changed from 4
+   const concurrentBatches = 2; // Changed from 4
    ```
 
 ### Change Batch Size
+
 To use larger batches (e.g., 10 indicators per batch):
 
 1. Update `.env`:
@@ -113,8 +120,8 @@ To use larger batches (e.g., 10 indicators per batch):
 
 2. Update scripts:
    ```typescript
-   const batchSize = 10;  // Changed from 5
-   const concurrentBatches = 2;  // Adjust to maintain total concurrency
+   const batchSize = 10; // Changed from 5
+   const concurrentBatches = 2; // Adjust to maintain total concurrency
    ```
 
 ## Monitoring
@@ -190,18 +197,22 @@ CREATE TABLE pipeline_stats (
 ## Troubleshooting
 
 ### "Too many concurrent requests"
+
 - Reduce `concurrentBatches` in scripts
 - Increase `INTERNAL_EMIT_DELAY_MS` in `.env`
 
 ### "Out of memory"
+
 - Reduce `concurrentBatches` (fewer indicators processing simultaneously)
 - Use smaller LLM model in LM Studio
 
 ### "Database locked"
+
 - SQLite handles concurrency well with WAL mode (enabled by default)
 - If issues persist, consider switching to PostgreSQL for production
 
 ### Batches not completing
+
 - Check logs for errors in individual steps
 - Query `processing_log` table for failed stages
 - Increase timeout in `waitForBatchCompletion()` if needed

apps/classify/workflow/README.md

@@ -378,6 +378,7 @@ Results are stored in the following Motia state groups:
 Deploy the workflow service to Railway for production-scale processing with horizontal scaling and PostgreSQL/TimescaleDB persistence.
 
 **Performance:**
+
 - **Local (M3)**: ~30-40 indicators/min
 - **Railway (3 replicas)**: ~150 indicators/min (5-6× faster)
 - **10,903 indicators**: ~73 minutes on Railway vs ~6 hours locally
@@ -439,12 +440,14 @@ NODE_ENV=production
 ### API Usage
 
 **Health Check:**
+
 ```bash
 curl https://your-service.up.railway.app/health
 # Response: {"status":"ok","timestamp":"...","service":"classify-workflow"}
 ```
 
 **Classify Batch:**
+
 ```bash
 curl -X POST https://your-service.up.railway.app/classify/batch \
   -H "Content-Type: application/json" \
@@ -457,23 +460,27 @@ curl -X POST https://your-service.up.railway.app/classify/batch \
 ### Rate Limits & Scaling
 
 **OpenAI GPT-4o-mini Tier 1:**
+
 - TPM: 200,000 tokens/minute
 - Per indicator: ~1,000 tokens (2-3 LLM calls)
 - Max throughput: ~200 indicators/minute
 
 **Recommended Configuration:**
+
 - 3 replicas × 50 concurrent each = 150 concurrent total
 - TPM usage: 75% (150K/200K)
 - 25% headroom for variance/retries
 
 **Scaling:**
+
 - 2 replicas: ~100 indicators/min (safe, 50% TPM)
 - 3 replicas: ~150 indicators/min (recommended, 75% TPM)
 - 4 replicas: ~200 indicators/min (max Tier 1, 100% TPM)
 
 ### Monitoring
 
 **Track Performance:**
+
 ```sql
 -- View batch statistics
 SELECT * FROM pipeline_stats ORDER BY batch_start_time DESC LIMIT 10;
@@ -489,23 +496,27 @@ FROM classifications;
 ```
 
 **Railway Metrics:**
+
 - Service health via `/health` endpoint
 - CPU/Memory usage in Railway dashboard
 - Request rate and latency
 
 ### Cost Analysis
 
 **API Costs (OpenAI GPT-4o-mini):**
+
 - Per indicator: ~$0.00382
 - 10,903 indicators: ~$42
 - Same cost regardless of replicas!
 
 **Infrastructure (Railway):**
+
 - Workflow service: ~$10-20/month (3 replicas)
 - Postgres: ~$10-20/month
 - Total: ~$20-40/month
 
 **Time Savings:**
+
 - Local: ~6 hours per 10.9K run
 - Railway: ~1.2 hours per 10.9K run
 - Saves: ~4.8 hours per run

apps/classify/workflow/SQLITE_MIGRATION_SUMMARY.md

@@ -12,6 +12,7 @@
 - **PostgreSQL:** Uses `$1`, `$2`, etc. placeholders
 
 **Files Updated:**
+
 - [src/db/repository.ts](./src/db/repository.ts) - All query methods now detect database type and use appropriate placeholders
 - [src/db/schema.ts](./src/db/schema.ts) - Added `metadata` column to `processing_log` table
 - [steps/classify-flow/complete-classify.step.ts](./steps/classify-flow/complete-classify.step.ts) - Fixed old `db.prepare()` calls, converted booleans to integers, JSON-stringified arrays/objects
@@ -21,6 +22,7 @@
 **Problem:** SQLite only accepts primitives (numbers, strings, bigints, buffers, null) but we were trying to bind JavaScript booleans and objects.
 
 **Solution:**
+
 - **Booleans → Integers:** All boolean fields (`is_cumulative`, `is_currency_denominated`, `boolean_review_passed`) now convert to 0/1
 - **Objects/Arrays → JSON Strings:** Fields like `boolean_review_fields_wrong` and `final_review_corrections` are JSON-stringified before saving
 
@@ -29,6 +31,7 @@
 **Problem:** `processing_log` table was missing the `metadata` column that was added to PostgreSQL.
 
 **Solution:**
+
 ```bash
 sqlite3 ./data/classify-workflow-local-dev.db "ALTER TABLE processing_log ADD COLUMN metadata TEXT;"
 ```
@@ -40,6 +43,7 @@ sqlite3 ./data/classify-workflow-local-dev.db "ALTER TABLE processing_log ADD CO
 **Solution:** Reduced to **1 concurrent batch** (5 indicators at a time):
 
 **Files Updated:**
+
 - [scripts/run-random.ts](./scripts/run-random.ts) - `concurrentBatches = 1`
 - [scripts/run-all.ts](./scripts/run-all.ts) - `concurrentBatches = 1`
 
@@ -48,6 +52,7 @@ sqlite3 ./data/classify-workflow-local-dev.db "ALTER TABLE processing_log ADD CO
 **Problem:** When indicators got stuck, the script would wait forever with no feedback.
 
 **Solution:** Added smart progress detection:
+
 - Detects when no progress for 20 seconds
 - Shows which indicators are stuck and at which stage
 - Shows error messages for failed indicators
@@ -56,6 +61,7 @@ sqlite3 ./data/classify-workflow-local-dev.db "ALTER TABLE processing_log ADD CO
 ## Final Configuration
 
 ### Environment Variables (`.env`)
+
 ```bash
 # SQLite Database
 CLASSIFY_DB=sqlite
@@ -67,20 +73,23 @@ INTERNAL_EMIT_DELAY_MS=500      # 500ms delay between batches
 ```
 
 ### Script Defaults
+
 ```typescript
-const batchSize = 5;              // 5 indicators per batch
-const concurrentBatches = 1;      // 1 batch at a time
+const batchSize = 5; // 5 indicators per batch
+const concurrentBatches = 1; // 1 batch at a time
 // Total concurrency: 5 indicators × ~6 LLM stages = ~30 API calls max
 ```
 
 ## Performance Characteristics
 
 ### Before
+
 - **Configuration:** 4 batches × 5 indicators = 20 concurrent
 - **LLM Calls:** ~120 concurrent API calls
 - **Result:** Rate limiting, stuck indicators, incomplete batches
 
 ### After
+
 - **Configuration:** 1 batch × 5 indicators = 5 concurrent
 - **LLM Calls:** ~30 concurrent API calls max
 - **Result:** Stable processing, all indicators complete, no rate limits
@@ -114,13 +123,15 @@ Next batch of 5 starts
 ## Database Compatibility
 
 ### SQLite (Local Development)
+
 - ✅ Proper `?` placeholders
 - ✅ Boolean values as 0/1 integers
 - ✅ JSON fields as TEXT
 - ✅ WAL mode enabled for concurrency
 - ✅ All schema columns present
 
 ### PostgreSQL (Production)
+
 - ✅ Proper `$1, $2` placeholders
 - ✅ Boolean values as BOOLEAN type
 - ✅ JSON fields as JSONB
@@ -130,11 +141,13 @@ Next batch of 5 starts
 ## Usage
 
 ### Run 50 Random Indicators
+
 ```bash
 deno task run:random -- -50 openai
 ```
 
 ### Expected Output
+
 ```
 🚀 Processing 50 indicators in 10 batches of 5...
    Provider: openai
@@ -152,19 +165,22 @@ deno task run:random -- -50 openai
 ```
 
 ### Average Timing
+
 - **Per indicator:** ~25-35 seconds (with OpenAI GPT-4.1-mini)
 - **Per batch (5 indicators):** ~30-45 seconds
 - **50 indicators total:** ~5-8 minutes
 
 ## Monitoring
 
 ### Check Completion Status
+
 ```bash
 sqlite3 ./data/classify-workflow-local-dev.db \
   "SELECT COUNT(*) FROM classifications;"
 ```
 
 ### Check Recent Activity
+
 ```bash
 sqlite3 ./data/classify-workflow-local-dev.db \
   "SELECT stage, status, COUNT(*) as count
@@ -174,6 +190,7 @@ sqlite3 ./data/classify-workflow-local-dev.db \
 ```
 
 ### Check Failed Indicators
+
 ```bash
 sqlite3 ./data/classify-workflow-local-dev.db \
   "SELECT indicator_id, stage, error_message
@@ -210,17 +227,20 @@ These are non-critical and logged as warnings. The pipeline will continue proces
 ## Next Steps
 
 ### To Increase Throughput (if no rate limits)
+
 1. Increase `concurrentBatches` to 2
 2. Monitor for stuck indicators
 3. Adjust based on API performance
 
 ### To Switch to PostgreSQL
+
 1. Set `POSTGRES_URL` environment variable
 2. Remove or comment out `CLASSIFY_DB=sqlite`
 3. Run migrations: `deno task migrate`
 4. Restart dev server
 
 ### To Use Local LLM (Free)
+
 1. Install LM Studio
 2. Load a model (e.g., Mistral 7B)
 3. Set environment: `LLM_PROVIDER=local`
@@ -229,22 +249,27 @@ These are non-critical and logged as warnings. The pipeline will continue proces
 ## Files Changed
 
 ### Database Layer
+
 - ✅ [src/db/repository.ts](./src/db/repository.ts)
 - ✅ [src/db/schema.ts](./src/db/schema.ts)
 - ✅ [src/db/client.ts](./src/db/client.ts)
 
 ### Workflow Steps
+
 - ✅ [steps/classify-flow/complete-classify.step.ts](./steps/classify-flow/complete-classify.step.ts)
 
 ### Scripts
+
 - ✅ [scripts/run-random.ts](./scripts/run-random.ts)
 - ✅ [scripts/run-all.ts](./scripts/run-all.ts)
 
 ### Configuration
+
 - ✅ [.env](./.env)
 - ✅ [.env.example](./.env.example)
 
 ### Documentation
+
 - ✅ [BATCHING.md](./BATCHING.md)
 - ✅ [examples/README.md](./examples/README.md)
 - ✅ [examples/parallel-batches.ts](./examples/parallel-batches.ts)