Feature Request: Display cost estimates in session_status

Summary

Add cost estimation to /status and session_status output based on local pricing config and LiteLLM database.

Problem

Currently, session_status shows token usage but no cost information:

🧮 Tokens: 84k in / 396 out

Users need to manually calculate costs or use external tools to track spending. This makes it difficult to:

• Monitor session costs in real-time
• Make informed decisions about model selection
• Track spending across conversations
• Set budget limits

Proposed Solution

Enhance session_status to display cost estimates when pricing data is available:

🧮 Tokens: 84k in / 396 out
💰 Cost: $0.252 in / $0.006 out ≈ $0.26 total

Implementation Approaches

Option 1: Provider-reported costs (when available)

• Use cost data from provider responses (OpenRouter, some APIs include this)
• Most accurate, but only works for supported providers

Option 2: Local config-based estimation

• Calculate from models.providers[].models[].cost in openclaw.json
• Works for all models if pricing is configured
• User-controlled, no API dependency

Option 3: LiteLLM database integration

• Use LiteLLM's pricing database (2479+ models, updated daily)
• GitHub: https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json
• Covers most major providers (OpenAI, Anthropic, Google, AWS, Azure, etc.)
• Can be cached locally, updated via cron

Recommended: Option 2 + 3 (local config as fallback, LiteLLM as primary source)

Use Cases

1. Real-time cost monitoring

User runs /status during a conversation to check current spend:

💰 Cost: $0.26 (84k tokens)

2. Model comparison

User compares costs before switching models:

Current: claude-sonnet-4-5 → $0.26
Alternative: gpt-4o → $0.19 (-27%)

3. Budget awareness

User sees cumulative costs approaching a limit:

Session cost: $2.45 / $5.00 daily budget

Configuration

Optional config to control display:

{
  "agents": {
    "defaults": {
      "showCostInStatus": true,  // default: true when pricing available
      "costPricingSource": "auto"  // auto|litellm|config|provider
    }
  }
}

Benefits

• Transparency: Users see costs immediately
• Cost awareness: Helps prevent surprise bills
• Model selection: Informed decisions about model tradeoffs
• Budget control: Foundation for future budget limits/warnings
• No API calls: Uses local data (config or cached LiteLLM pricing)

Implementation Notes

• Cost should be marked as "estimate" when calculated locally
• Show "Cost: unavailable" if no pricing data exists
• Include cache read/write costs when supported
• Format with currency symbol and appropriate precision ($0.0234)

Related Issues

• Feature: Expose OpenRouter usage cost to agent runtime #9016 - Expose OpenRouter usage cost to agent runtime (provider-specific)
• This feature request is broader: applies to all providers via local estimation

Additional Context

Users have already built workarounds:

• External cost tracking tools (e.g., clawdbot-cost-monitor on GitHub)
• Custom scripts that parse session data and calculate costs
• Manual tracking in spreadsheets

Native support would improve UX and reduce friction for cost-conscious users.