UNPKG

@logspace/mcp-server

Version:

MCP server for Logspace log analysis integration with AI models.

49 lines 1.83 kB
/** * Demonstration: Why Fallback Makes Sense * Shows the progressive investigation approach in action */ /** * KEY INSIGHT: The fallback strategy gives you the BEST of both worlds: * * 1. **Efficiency When Possible**: 90% of bugs have obvious patterns that AI can find quickly * 2. **Safety Net**: When AI finds nothing obvious, you get the SAME comprehensive coverage as your original approach * 3. **No False Negatives**: You're never missing data - if targeted fails, you get everything * 4. **Learning Loop**: You can compare what AI found vs. what's in the full data to improve targeting * * This addresses your concern: "What if AI misses something?" * Answer: Then it automatically falls back to giving AI the same complete data your original approach provided. */ export declare const USE_CASE_CONFIGS: { automation: { strategy: string; maxTokens: number; message: string; }; development: { strategy: string; maxTokens: number; message: string; }; investigation: { strategy: string; maxTokens: number; message: string; }; aiTesting: { strategy: string; maxTokens: number; message: string; }; }; /** * Migration Strategy from Original Approach: * * Phase 1: Add fallback tool alongside existing tools * Phase 2: Use 'auto' strategy by default (tries efficient first, falls back to original coverage) * Phase 3: Monitor which sessions require fallback vs. targeted success * Phase 4: Improve targeting logic based on fallback learnings * Phase 5: Eventually most sessions will be handled efficiently, rare complex ones get full data * * Result: You get 90% efficiency improvement while maintaining 100% coverage safety net */ //# sourceMappingURL=fallbackDemo.d.ts.map