Add cost tracking and centralized logging; propagate LLM operation metadata

- Introduce cost tracking system and CLI:
  - Add brain_core/cost_tracker.py for recording/aggregating LLM usage
  - Add brain_cli/cost_commands.py to view summary, trends, estimate, breakdown, export, pricing
  - Add cost.json example data
  - Update .env.example and README with cost/logging/pricing options and new commands

- Centralized logging:
  - Add brain_core/logging_config.py with setup_logging, helpers, and LLM/timing log functions
  - Wire setup_logging into brain_cli/main.py and add --verbose/--debug flags + version callback

- LLM usage metadata and cost recording:
  - Extend LLMClient.generate_sync signature to accept operation and entry_date
  - AzureOpenAIClient now records token usage, estimates cost via CostTracker, and calls log_llm_call
  - Propagate operation/entry_date parameters in template_generator, llm_analysis, report_generator, plan_commands

- Behavior and config tweaks:
  - Update brain_core/config.py to expose cost/logging config
  - Update constants: DAILY_PROMPT_COUNT=2, PROMPT_CONTEXT_DAYS=2, add AZURE_OPENAI_PRICING map
  - Lower-level logging swapped to debug for LLM/prompt generation messages

- Tests:
  - Update tests to use create_autospec(LLMClient) and assert operation/entry_date are passed for task extraction

This brings end-to-end cost tracking, richer logging, and consistent propagation of LLM call metadata across the codebase.

Author: madebygps

.env.example

@@ -20,7 +20,20 @@ AZURE_SEARCH_ENDPOINT=https://your-search-service.search.windows.net
 AZURE_SEARCH_API_KEY=your-search-api-key-here
 AZURE_SEARCH_INDEX_NAME=second-brain-notes
 
-# Optional: Daemon configuration
-DAEMON_AUTO_LINK_TIME=23:00
-DAEMON_WEEKLY_ANALYSIS=true
-DAEMON_REFRESH_DAYS=30
+# Optional: Cost Tracking Configuration
+BRAIN_COST_DB_PATH=~/.brain/costs.db
+
+# Optional: Logging Configuration
+BRAIN_LOG_LEVEL=INFO
+BRAIN_LOG_FILE=~/.brain/logs/brain.log
+
+# Optional: Custom Pricing (override Azure OpenAI pricing)
+# Prices are per 1K tokens
+AZURE_GPT4O_INPUT_PRICE=0.03
+AZURE_GPT4O_OUTPUT_PRICE=0.06
+AZURE_GPT4O_MINI_INPUT_PRICE=0.0015
+AZURE_GPT4O_MINI_OUTPUT_PRICE=0.006
+AZURE_GPT4_INPUT_PRICE=0.03
+AZURE_GPT4_OUTPUT_PRICE=0.06
+AZURE_GPT35_TURBO_INPUT_PRICE=0.0015
+AZURE_GPT35_TURBO_OUTPUT_PRICE=0.002

README.md

@@ -23,6 +23,20 @@ AI-powered journaling with semantic backlinks and intelligent book notes search.
 - Rich formatted results
 - **Requires Azure AI Search** (no local alternative)
 
+**Cost Tracking:**
+- Real-time Azure OpenAI usage tracking
+- Local SQLite database storage
+- Cost summaries, trends, and projections
+- Per-operation breakdowns (diary, planning, analysis)
+- Export to CSV for billing analysis
+
+**Logging:**
+- Rich-formatted colored console output
+- Clean UX with smart log suppression (no interruptions during spinners)
+- Multiple verbosity levels (--verbose, --debug)
+- File and console output options
+- Suppressed noisy third-party HTTP logs
+
 **LLM Support:**
 - Azure OpenAI - Full functionality for all commands
 
@@ -68,9 +82,23 @@ uv run brain notes search "topic"     # Search book notes
 uv run brain notes search "topic" --semantic --detailed
 uv run brain notes status             # Check connection
 
-
+# Cost tracking and analysis
+uv run brain cost summary             # Usage summary for last 30 days
+uv run brain cost trends 30           # Daily cost trends
+uv run brain cost estimate            # Monthly cost projection
+uv run brain cost breakdown           # Per-operation breakdown
+uv run brain cost export data.json    # Export to JSON
+uv run brain cost pricing             # Show pricing
+
+# Logging (Rich-formatted, colored output)
+uv run brain --verbose <command>      # Show key operations
+uv run brain --debug <command>        # Full diagnostics with LLM details
 ```
 
+> **Clean UX:** By default, logs are suppressed during operations to avoid interrupting spinners. Use `--verbose` or `--debug` for detailed information with beautiful Rich formatting.
+
+> **Cost Tracking:** All Azure OpenAI usage is automatically tracked in a local SQLite database (`~/.brain/costs.db`). Data stays private and grows ~10-50 MB/year for typical use.
+
 ## Configuration
 
 **Required Paths:**
@@ -88,6 +116,21 @@ uv run brain notes status             # Check connection
 - `AZURE_SEARCH_API_KEY` - Azure AI Search API key
 - `AZURE_SEARCH_INDEX_NAME` - Search index name (default: second-brain-notes)
 
+**Cost Tracking** (optional configuration):
+- `BRAIN_COST_DB_PATH` - Path to cost tracking database (default: ~/.brain/costs.db)
+- `BRAIN_LOG_LEVEL` - Logging level: DEBUG, INFO, WARNING, ERROR (default: INFO)
+- `BRAIN_LOG_FILE` - Path to log file (optional, logs to console if not set)
+
+**Custom Pricing** (optional - override Azure OpenAI pricing):
+- `AZURE_GPT4O_INPUT_PRICE` - Price per 1K input tokens for gpt-4o (default: 0.03)
+- `AZURE_GPT4O_OUTPUT_PRICE` - Price per 1K output tokens for gpt-4o (default: 0.06)
+- `AZURE_GPT4O_MINI_INPUT_PRICE` - Price per 1K input tokens for gpt-4o-mini (default: 0.0015)
+- `AZURE_GPT4O_MINI_OUTPUT_PRICE` - Price per 1K output tokens for gpt-4o-mini (default: 0.006)
+- `AZURE_GPT4_INPUT_PRICE` - Price per 1K input tokens for gpt-4 (default: 0.03)
+- `AZURE_GPT4_OUTPUT_PRICE` - Price per 1K output tokens for gpt-4 (default: 0.06)
+- `AZURE_GPT35_TURBO_INPUT_PRICE` - Price per 1K input tokens for gpt-35-turbo (default: 0.0015)
+- `AZURE_GPT35_TURBO_OUTPUT_PRICE` - Price per 1K output tokens for gpt-35-turbo (default: 0.002)
+
 ## Entry Format
 
 **Morning Plan Entry** (`YYYY-MM-DD-plan.md`):
@@ -147,6 +190,8 @@ uv run pytest --cov              # Run with coverage
 uv run pytest tests/brain_core/  # Run specific module
 ```
 
+**Coverage**: 74 tests with 45% overall coverage, 96% on critical paths (plan_commands.py)
+
 ## License
 
 MIT