adaptive-memory-multi-model-router 2.15.3 → 2.15.4

package/OPPORTUNITIES_100.md

@@ -1,277 +0,0 @@
-# A3M Router: 100+ Integration Opportunities
-
-## Executive Summary
-
-This document identifies 100+ opportunities for A3M Router integration across major software categories. Each opportunity represents a tool, framework, or platform that either lacks LLM routing capabilities or currently relies on LiteLLM (which has CVEs and is being abandoned).
-
-**Key Insight:** browser-use removed LiteLLM due to CVE-2026-42271. This creates a cascading opportunity across the entire AI tooling ecosystem.
-
----
-
-## Category 1: AI Agent Frameworks (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 1 | **LangChain/LangGraph** | 35k | Multi-provider but no intelligent routing | First-class A3M integration |
-| 2 | **CrewAI** | 22k | Uses LiteLLM for routing | Direct replacement for LiteLLM |
-| 3 | **AutoGen** | 35k | Multi-model but manual selection | A3M auto-routing layer |
-| 4 | **LlamaIndex** | 35k | Basic model selection | Enhanced routing for RAG |
-| 5 | **Microsoft Semantic Kernel** | 9k | Enterprise-focused, expensive | Cost optimization routing |
-| 6 | **Haystack** | 15k | Limited routing | Pipeline integration |
-| 7 | **Botpress** | 12k | Rule-based routing | AI-powered routing |
-| 8 | **Rasa** | 12k | Custom NLU models | Enterprise routing |
-| 9 | **Dialogflow** | 8k | Google's model only | Multi-provider routing |
-| 10 | **OpenAI Agents SDK** | 5k | OpenAI-only | Multi-provider alternative |
-
-**Key Finding:** "Best 5 OpenAI Agents SDK Alternatives 2026" article highlights multi-provider routing as the #1 requested feature.
-
----
-
-## Category 2: Browser & Web Automation (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 11 | **Playwright** | 65k | No built-in LLM | A3M-powered automation |
-| 12 | **Selenium** | 30k | No LLM integration | AI-powered testing |
-| 13 | **Puppeteer** | 85k | No LLM integration | Chrome automation + LLM |
-| 14 | **Cypress** | 45k | No AI features | Intelligent test generation |
-| 15 | **TestCafe** | 10k | No AI features | AI-powered E2E testing |
-| 16 | **Robot Framework** | 12k | Keyword-based | LLM-driven automation |
-| 17 | **Cheerio** | 28k | Scraping only | AI content extraction |
-| 18 | **Crawlee** | 22k | Actor framework | Smart crawling with routing |
-| 19 | **Apify** | 15k | Pre-built scrapers | LLM-optimized extraction |
-| 20 | **FireCrawl** | 18k | ML-powered crawling | Enhanced routing |
-
-**Key Finding:** browser-use PR submitted (#5378). More tools need similar integration.
-
----
-
-## Category 3: LLM Serving & Inference (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 21 | **vLLM** | 45k | Single model serving | Multi-model gateway |
-| 22 | **Ollama** | 120k | Local models only | Remote model routing |
-| 23 | **LocalAI** | 25k | Self-hosted LLMs | Multi-provider gateway |
-| 24 | **Text Generation Inference** | 18k | HuggingFace serving | Intelligent routing layer |
-| 25 | **Ray Serve** | 8k | Distributed serving | Model selection |
-| 26 | **LM Studio** | 35k | Desktop app | API gateway integration |
-| 27 | **Ollama Web UI** | 12k | Ollama frontend | Routing backend |
-| 28 | **Jan** | 15k | Local AI interface | Multi-model routing |
-| 29 | **SGLang** | 8k | Structured generation | Routing optimization |
-| 30 | **mlc-llm** | 18k | Local deployment | Cloud routing hybrid |
-
-**Key Finding:** "vLLM vs Ollama vs LM Studio" comparisons all miss intelligent routing.
-
----
-
-## Category 4: RAG & Vector Databases (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 31 | **Chroma** | 18k | No routing | Query routing optimization |
-| 32 | **Pinecone** | 8k | Managed vectors | Multi-model RAG |
-| 33 | **Weaviate** | 12k | Hybrid search | LLM selection layer |
-| 34 | **Qdrant** | 22k | Vector search | Intelligent reranking |
-| 35 | **Milvus** | 28k | Scale vectors | Multi-model queries |
-| 36 | **LlamaIndex RAG** | Built-in | Basic selection | Advanced routing |
-| 37 | **LangChain RAG** | Built-in | Model picker | A3M-powered routing |
-| 38 | **Elasticsearch** | 65k | BM25 search | AI-powered retrieval |
-| 39 | **Typesense** | 22k | Typo tolerance | LLM-enhanced search |
-| 40 | **DocArray** | 8k | Vector indexing | Multi-model embedding |
-
-**Key Finding:** RAG pipelines spend 60-80% on LLM calls. Routing can cut costs 90%.
-
----
-
-## Category 5: No-Code & Low-Code Platforms (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 41 | **Zapier** | 8k | 5000+ integrations | AI action routing |
-| 42 | **Make (Integromat)** | 6k | Visual automation | LLM decision routing |
-| 43 | **n8n** | 55k | Self-hosted automation | Built-in routing |
-| 44 | **Bubble** | 25k | No-code app builder | AI integration layer |
-| 45 | **Adalo** | 8k | No-code mobile | LLM backend |
-| 46 | **Glide** | 12k | Spreadsheet to app | AI features |
-| 47 | **Softr** | 15k | Airtable front-end | AI enhancement |
-| 48 | **Stacker** | 5k | Notion-based apps | LLM integration |
-| 49 | **FlutterFlow** | 18k | No-code Flutter | AI code generation |
-| 50 | **Retool** | 28k | Internal tools | AI action routing |
-
-**Key Finding:** n8n has 55k stars and growing AI features. Integration opportunity exists.
-
----
-
-## Category 6: RPA & Business Automation (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 51 | **UiPath** | 12k | Enterprise RPA | AI-powered automation |
-| 52 | **Automation Anywhere** | 8k | RPA leader | LLM integration |
-| 53 | **Power Automate** | 6k | Microsoft ecosystem | Azure OpenAI routing |
-| 54 | **Blue Prism** | 5k | Legacy RPA | AI upgrade path |
-| 55 | **WorkFusion** | 4k | Intelligent automation | Multi-model routing |
-| 56 | **Automation Edge** | 3k | IT automation | LLM decision routing |
-| 57 | **Kofax** | 5k | Document processing | AI-powered OCR |
-| 58 | **WinAutomation** | 3k | Windows RPA | AI enhancement |
-| 59 | **Jidoka** | 2k | DevOps automation | Intelligent routing |
-| 60 | **Windmill** | 8k | Script automation | LLM-native scripts |
-
-**Key Finding:** RPA tools are adding AI but lack cost optimization. Opportunity for routing.
-
----
-
-## Category 7: Job & Career Automation (10 opportunities)
-
-| # | Tool | Type | Current LLM Setup | Opportunity |
-|---|------|------|-------------------|-------------|
-| 61 | **Simplify Copilot** | Browser ext | GPT-4 only | Multi-model application |
-| 62 | **LazyApply** | Job applier | Basic AI | Smart model selection |
-| 63 | **AngelOne Job** | Trading jobs | ML only | AI career coach |
-| 64 | **Applyai** | Job automation | GPT dependent | Cost optimization |
-| 65 | **WonsultAI** | Resume builder | Single model | Multi-tier writing |
-| 66 | **Resumeworded** | Resume checker | Basic AI | Enhanced evaluation |
-| 67 | **Jobscan** | Resume optimizer | Keyword match | AI-powered matching |
-| 68 | **Kickresume** | Resume builder | GPT integration | Multi-model output |
-| 69 | **Rezi** | Resume AI | Single AI | Feature routing |
-| 70 | **Hiration** | Career platform | Multi-tool | Unified routing |
-
-**Key Finding:** Job automation tools spend heavily on GPT-4. Routing can cut costs 95%.
-
----
-
-## Category 8: Customer Support & Chatbots (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 71 | **Intercom Fin** | 8k | GPT-4 only | Multi-model support |
-| 72 | **Zendesk AI** | 6k | OpenAI backend | Routing layer |
-| 73 | **Freshdesk Freddy** | 5k | AI assist | Cost optimization |
-| 74 | **Drift** | 4k | Rule-based + AI | Intelligent routing |
-| 75 | **Ada** | 6k | Automated support | Multi-tier responses |
-| 76 | **Forethought** | 3k | Support automation | LLM selection |
-| 77 | **Kore.ai** | 5k | Enterprise chatbot | Multi-model routing |
-| 78 | **Yellow.ai** | 8k | Omnichannel AI | Cost optimization |
-| 79 | **LiveChat** | 4k | Chat platform | AI routing |
-| 80 | **Chatfuel** | 6k | Messenger bots | AI enhancement |
-
-**Key Finding:** Customer support AI costs spiral with GPT-4. Routing can maintain quality at 10% cost.
-
----
-
-## Category 9: Code Generation & Development Tools (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 81 | **GitHub Copilot** | 15k | GPT-4 / Codex | Enterprise routing |
-| 82 | **Cursor** | 45k | Multiple models | Smart model selection |
-| 83 | **Claude Code** | 20k | Claude only | Multi-model ensemble |
-| 84 | **Tabnine** | 12k | Local + cloud | Optimal model routing |
-| 85 | **Amazon CodeWhisperer** | 8k | Amazon models | Multi-provider |
-| 86 | **Sourcegraph Cody** | 18k | Claude + GPT | Intelligent routing |
-| 87 | **Codeium** | 25k | Free AI coding | Premium routing tier |
-| 88 | **CodeRabbit** | 8k | PR review AI | Multi-model review |
-| 89 | **Mutable AI** | 6k | Production coding | Adaptive model selection |
-| 90 | **Seek** | 5k | Code search | LLM-powered search |
-
-**Key Finding:** "Claude Code Router vs LiteLLM" articles show demand for cost control in coding tools.
-
----
-
-## Category 10: Data & Analytics Platforms (10 opportunities)
-
-| # | Tool | Stars | Current LLM Setup | Opportunity |
-|---|------|-------|-------------------|-------------|
-| 91 | **Metabase** | 35k | SQL + AI | Natural language queries |
-| 92 | **Looker** | 8k | Embedded analytics | AI query routing |
-| 93 | **Mode Analytics** | 12k | Python/R notebooks | LLM-powered analysis |
-| 94 | **Superset** | 65k | Apache project | AI chart generation |
-| 95 | **Redash** | 25k | Query editor | Intelligent visualization |
-| 96 | **Chartbrew** | 3k | Data visualization | AI-powered dashboards |
-| 97 | **Grafana** | 65k | Observability | LLM alerting |
-| 98 | **Hex** | 8k | Data notebooks | AI-powered exploration |
-| 99 | **Observable** | 5k | Visual analytics | LLM integration |
-| 100 | **Domo** | 4k | Cloud BI | AI-powered insights |
-
-**Key Finding:** BI tools adding AI but pricing is opaque. Routing provides cost transparency.
-
----
-
-## Category 11: Additional Opportunities (15 opportunities)
-
-| # | Tool/Category | Type | Opportunity |
-|---|---------------|------|-------------|
-| 101 | **WordPress AI** | CMS | Jetpack AI, Elementor AI optimization |
-| 102 | **Shopify AI** | E-commerce | Storefront AI, customer support routing |
-| 103 | **Salesforce Einstein** | CRM | Multi-model routing for CRM AI |
-| 104 | **HubSpot AI** | Marketing | Content and support optimization |
-| 105 | **Notion AI** | Productivity | Multi-model writing assistant |
-| 106 | **Slack AI** | Communication | Enterprise LLM routing |
-| 107 | **Figma AI** | Design | Design generation routing |
-| 108 | **Canva AI** | Design | Content generation optimization |
-| 109 | **Adobe Firefly** | Creative | Multi-model creative AI |
-| 110 | **Jira AI** | Project | Intelligent issue routing |
-| 111 | **Linear AI** | Dev management | Smart issue classification |
-| 112 | **Airtable AI** | Database | Formulas and automation |
-| 113 | **Coda AI** | Documents | Intelligent docs |
-| 114 | **Monday.com AI** | PM | Work management optimization |
-| 115 | **Asana AI** | Task management | Smart task routing |
-
----
-
-## Priority Actions
-
-### Tier 1: Quick Wins (Already started)
-1. ✅ browser-use PR #5378 - 108k stars, massive visibility
-2. 🔄 LangChain integration - First-class status
-3. 🔄 LlamaIndex integration - Already documented
-
-### Tier 2: High-Impact (Next Sprint)
-- CrewAI: 22k stars, uses LiteLLM
-- n8n: 55k stars, needs built-in routing
-- Cursor: 45k stars, multi-model support
-- AutoGen: 35k stars, enterprise focus
-
-### Tier 3: Strategic Plays
-- GitHub Copilot enterprise routing
-- Salesforce/HubSpot CRM integration
-- WordPress/Jetpack AI optimization
-
----
-
-## Competitive Positioning Matrix
-
-| Competitor | Weakness | A3M Advantage |
-|------------|----------|----------------|
-| LiteLLM | CVE-2026-42271, supply chain attack | Secure, lightweight |
-| OpenRouter | Expensive, limited routing | 90% cheaper routing |
-| Vercel AI | Vendor lock-in | Provider agnostic |
-| Portkey | Enterprise pricing | Open source, free |
-| Braintrust | Eval focus, not routing | Pure routing optimization |
-
----
-
-## Success Metrics
-
-Target integrations by category:
-- **10k+ stars tools**: 15 integrations
-- **25k+ stars tools**: 8 integrations
-- **50k+ stars tools**: 5 integrations
-- **100k+ stars tools**: 2 integrations
-
-Total potential visibility: **500k+ GitHub stars** of tools with A3M integration
-
----
-
-## Next Steps
-
-1. **This week**: Merge browser-use PR, announce on HN
-2. **This month**: 5 PRs to high-star tools (CrewAI, n8n, Cursor, AutoGen, LangChain)
-3. **This quarter**: 20 integrations across categories
-4. **This year**: 100+ integrations, become default routing layer
-
----
-
-*Generated: 2026-08-03*
-*Source: Web research, GitHub analysis, market positioning*

package/TODO_BROWSER_AUTOMATION.md

@@ -1,87 +0,0 @@
-# A3M Browser Automation - Remaining Tasks
-
-## Completed ✅
-- [x] browser-use integration (README, index.ts, package.json, examples)
-- [x] MCP server for browser tools (README, src/index.ts, package.json)
-- [x] sota-browser integration (README, package.json)
-- [x] Updated main README with browser automation positioning
-- [x] Updated AGENT_COUNCIL_FINDINGS.md
-
-## High Priority - Do Today
-
-### 1. Submit PR to browser-use
-**Why:** browser-use (108K stars) recently removed LiteLLM due to CVE. They're looking for alternatives.
-**How:**
-1. Fork browser-use
-2. Add `A3MRouter` to their supported LLM providers list
-3. Create example: `examples/a3m-router.ts`
-4. Submit PR with message highlighting:
-   - 70-95% cost savings
-   - Built-in stealth mode
-   - Parallel ensemble for reliability
-
-### 2. Publish MCP server to NPM
-```bash
-cd ~/a3m-router/integrations/mcp-browser
-npm publish
-```
-**Then:** Update docs to show one-command install
-
-### 3. Create Demo Video/GIF
-**What to record:**
-- Terminal showing A3M routing for form filling
-- Cost comparison output
-- browser-use + A3M demo
-
-**Tools:** terminalizer, asciinema, or OBS
-
-## Medium Priority - This Week
-
-### 4. Build Benchmark
-Run 100 form-filling tasks and compare:
-- A3M vs GPT-4o vs LiteLLM
-- Document cost savings
-- Publish to Hacker News
-
-### 5. GitHub Actions CI
-Add tests for integration packages:
-```yaml
-- name: Test browser-use integration
-  run: npm test -- integrations/browser-use
-```
-
-### 6. Discord/Community Engagement
-- Join browser-use Discord
-- Answer questions about A3M
-- Contribute to discussions
-
-## Low Priority - This Month
-
-### 7. Write Tutorial
-Create blog post: "How to Build a Cost-Optimized Job Application Bot with A3M + browser-use"
-
-### 8. Add More Examples
-- Web scraping with A3M
-- Data extraction pipeline
-- Multi-step form automation
-
-### 9. Monitor & Iterate
-- Track NPM downloads
-- Respond to GitHub issues
-- Update based on feedback
-
-## Success Metrics
-
-| Metric | Target (1 month) | Target (3 months) |
-|--------|-------------------|-------------------|
-| GitHub Stars | +50 | +200 |
-| NPM Downloads | 2K/week | 5K/week |
-| browser-use PR | Merged | Active use |
-| Community mentions | 5 | 20 |
-
-## Key Resources
-
-- **browser-use repo:** https://github.com/browser-use/browser-use
-- **browser-use integrations:** https://github.com/browser-use/browser-use/tree/main/browser_use/llm
-- **MCP Protocol:** https://modelcontextprotocol.io
-- **sota-browser:** ~/omniclaw/skills/browser/sota-browser

package/a3m-integrations-summary.md

@@ -1,73 +0,0 @@
-# A3M Router Integration Summary
-
-## Repositories Created This Session
-
-| # | Repo | Description | Stars | Status |
-|---|------|-------------|-------|--------|
-| 1 | [browser-use](https://github.com/browser-use/browser-use/pull/5378) | Browser automation | 108k | PR Open |
-| 2 | [crewAI](https://github.com/crewAIInc/crewAI/pull/6794) | Multi-agent framework | 57k | PR Open |
-| 3 | [n8n-nodes-a3m-router](https://github.com/Das-rebel/n8n-nodes-a3m-router) | n8n workflow automation | 55k | Pushed |
-| 4 | [a3m-autogen](https://github.com/Das-rebel/a3m-autogen) | AutoGen examples | 35k | Pushed |
-| 5 | [a3m-cursor-mcp](https://github.com/Das-rebel/a3m-cursor-mcp) | Cursor MCP server | 45k | Pushed |
-| 6 | [a3m-llamaindex](https://github.com/Das-rebel/a3m-llamaindex) | LlamaIndex RAG | 35k | Pushed |
-
-## Total Potential Impact
-
-- **Total stars of integrated tools**: 370k+
-- **PRs submitted**: 2 (browser-use, crewAI)
-- **Repositories created**: 4 new
-
-## Value Proposition
-
-### For Tool Users
-- 70-95% cost savings on LLM calls
-- Automatic model selection
-- No configuration needed
-- Built-in fallback handling
-
-### For Tool Maintainers
-- Differentiation from LiteLLM (CVE-free)
-- Value-add for users
-- Easy integration (OpenAI-compatible)
-
-## Next Steps
-
-### High Priority
-1. Get browser-use PR merged
-2. Get crewAI PR merged
-3. Create n8n community node submission
-
-### Medium Priority
-4. AutoGen official integration
-5. LlamaIndex official integration
-6. LangChain enhanced docs
-
-### Lower Priority
-7. RPA tool integrations (UiPath, Power Automate)
-8. Enterprise integrations (Salesforce, HubSpot)
-9. More framework integrations
-
-## Key Markets
-
-1. **AI Agent Frameworks**: 150k+ combined stars
-2. **Browser Automation**: 100k+ combined stars
-3. **Workflow Automation**: 55k+ stars
-4. **Coding Tools**: 45k+ stars
-5. **RAG Platforms**: 35k+ stars
-
-## Competitive Advantage
-
-| Feature | LiteLLM | A3M Router |
-|---------|---------|-------------|
-| Security | CVE-2026-42271 | Clean |
-| Size | Heavy | Lightweight |
-| Routing | Heuristic | Biology-inspired |
-| Cost | Markup | Provider rate |
-| Maintenance | Slow fixes | Fast |
-
-## Resources
-
-- [A3M Router](https://github.com/Das-rebel/a3m-router)
-- [Integration Progress](./INTEGRATION_PROGRESS.md)
-- [Opportunities List](./OPPORTUNITIES_100.md)
-

package/dist/routing/shadowSampler.d.ts

@@ -1,187 +0,0 @@
-/**
- * ShadowSampler — Value-Proportional Shadow Verification
- *
- * Implements Optimal Defense Theory (Rhoades 1979; McKey 1974; Zangerl & Bazzaz 1992)
- * from plant ecological economics:
- *
- *   "A plant allocates defensive compounds (defense cost) in proportion to
- *    tissue value, attack probability, and the marginal cost of defense."
- *
- * Mapped to A3M routing:
- *   - Tissue value     → query stakes / expected cost of wrong answer
- *   - Attack probability → probability that the primary provider fails/misbehaves
- *   - Defense cost    → cost of the shadow verification call
- *   - Marginal return → expected reduction in error rate from verification
- *
- * Instead of binary always-on (wasteful) or always-off (risky), ODT says:
- *   Sample shadow verification PROPORTIONALLY to expected value of verification.
- *
- * The sampling probability is:
- *   P(shadow) = f(expected_error_cost_reduction, query_risk)
- *
- * This is NOT the Zahavi "shared expert" (which always runs on 100% of queries).
- * ODT shadow runs probabilistically — only on queries where the expected
- * verification benefit exceeds the verification cost.
- *
- * Usage:
- *   const sampler = new ShadowSampler();
- *   const decision = await sampler.routeWithShadow(query, { strategy: 'auto' });
- *   // decision.hasShadow === true iff we sampled a shadow provider
- *   // decision.primaryResult and decision.shadowResult are compared automatically
- */
-import { extractQueryFeatures, RouteDecision } from './advancedRouter';
-export interface ShadowSamplerConfig {
-    /**
-     * Minimum query stakes (estimated cost of wrong answer) to consider shadow.
-     * Below this, P(shadow) = 0 regardless of other factors.
-     * In dollars — what a wrong answer costs you.
-     * Default: $0.01 (1 cent — roughly equivalent to a simple API call cost)
-     */
-    minQueryStake?: number;
-    /**
-     * Maximum probability of shadow verification.
-     * Set to 1.0 for always-on (Zahavi-style, expensive).
-     * Default: 0.15 (15% — ODT-proportional sampling)
-     */
-    maxShadowProbability?: number;
-    /**
-     * Risk threshold (from QueryFeatures.risk_profile) above which
-     * shadow probability scales up linearly.
-     * 0.0 = no risk scaling (all queries get same P(shadow))
-     * 0.5 = medium risk scales P(shadow) by 50%
-     * 1.0 = high risk scales P(shadow) by 100%
-     */
-    riskScalingFactor?: number;
-    /**
-     * Cost scaling factor — if query is expensive (large output),
-     * shadow is more worthwhile (higher P(shadow)).
-     * 0.0 = no cost scaling.
-     * Default: 0.3 (expensive queries get moderate boost in P(shadow))
-     */
-    costScalingFactor?: number;
-    /**
-     * Provider to use as shadow (cheapest reliable provider).
-     * If not specified, auto-selected from available providers.
-     */
-    shadowProvider?: string;
-    /**
-     * Always use this shadow provider (overrides auto-selection).
-     */
-    forceShadowProvider?: boolean;
-    /**
-     * If true, use complexity as a signal for verification value.
-     * Complex queries (code, math, reasoning) benefit more from verification.
-     * Default: true
-     */
-    useComplexitySignal?: boolean;
-    /**
-     * Debug mode: always shadow regardless of sampling.
-     * Default: false
-     */
-    forceShadow?: boolean;
-}
-export interface ShadowDecision extends RouteDecision {
-    /** Whether a shadow provider was sampled for this query */
-    hasShadow: boolean;
-    /** The sampled shadow provider (null if no shadow) */
-    shadowProvider: string | null;
-    /** Shadow sampling probability that was used */
-    shadowProbability: number;
-    /** The query features that triggered shadow decision */
-    features: ReturnType<typeof extractQueryFeatures>;
-    /** ODT rationale for the shadow decision */
-    shadowReasoning: string;
-}
-export interface ShadowedResponse {
-    primary: string;
-    shadow: string | null;
-    winner: 'primary' | 'shadow' | 'tie' | 'no-shadow';
-    confidenceDelta: number;
-}
-/**
- * Optimal Defense Theory Shadow Sampler
- *
- * Allocates shadow verification proportionally to query value and risk.
- *
- * The ODT sampling probability:
- *   P_shadow = min(maxP, baseP + risk_adj + cost_adj + complexity_adj)
- *
- * where:
- *   baseP     = baseline verification rate (ODT " constitutive defense")
- *   risk_adj  = risk_profile scaling (ODT "induced defense")
- *   cost_adj  = output cost scaling (value of correct answer)
- *   complexity_adj = query complexity signal
- */
-export declare class ShadowSampler {
-    private config;
-    private _shadowProvider;
-    private _shadowCount;
-    private _totalCount;
-    constructor(config?: ShadowSamplerConfig);
-    /**
-     * Auto-select the cheapest available provider as the shadow.
-     * Excludes the primary provider to ensure diversity.
-     */
-    private selectShadowProvider;
-    /**
-     * Estimate the query "stakes" — the expected cost of a wrong answer.
-     *
-     * ODT maps this to "tissue value": how much is this asset worth protecting?
-     * In routing terms: if this query fails, how much does it cost?
-     *
-     * We approximate this as:
-     *   stake ≈ estimated_output_tokens × cost_per_token × criticality_multiplier
-     *
-     * where criticality is derived from query complexity and risk_profile.
-     */
-    private estimateQueryStake;
-    /**
-     * Compute the ODT sampling probability for this query.
-     *
-     * ODT principle: defense (shadow) allocation ∝ expected benefit of defense.
-     * Expected benefit = P(failure) × cost_of_failure
-     *
-     * So P(shadow) scales with:
-     *   1. Query stake (expected cost of wrong answer)
-     *   2. Risk profile (probability of primary failure)
-     *   3. Output cost (verification ROI — cheaper outputs need less verification)
-     *   4. Complexity (complex queries benefit more from verification)
-     */
-    private computeShadowProbability;
-    /**
-     * Route a query, with optional ODT-proportional shadow verification.
-     *
-     * @param prompt - The user query
-     * @param options - Routing options (same as routeQuery)
-     * @returns ShadowDecision with shadow metadata
-     */
-    routeWithShadow(prompt: string, options?: {
-        available_models?: string[];
-        budget_multiplier?: number;
-    }): ShadowDecision;
-    /**
-     * Compare primary and shadow outputs.
-     * Returns the "better" answer and confidence delta.
-     *
-     * For production use: this would call both providers in parallel
-     * and compare outputs. For now, returns a stub that signals
-     * the caller should handle comparison.
-     */
-    compareOutputs(primaryAnswer: string, shadowAnswer: string | null): Promise<ShadowedResponse>;
-    /**
-     * Get sampling statistics for monitoring.
-     */
-    getStats(): {
-        shadowCount: number;
-        totalCount: number;
-        shadowRate: number;
-    };
-    /**
-     * Reset statistics counters.
-     */
-    resetStats(): void;
-    /**
-     * Update configuration at runtime.
-     */
-    configure(config: Partial<ShadowSamplerConfig>): void;
-}