@goodandready/dsh-moa 0.2.2 → 0.2.4

package/README.md

@@ -37,6 +37,10 @@ Single-model AI generation often suffers from blind spots, single-perspective bi
 4. **Instant Live Canvas Previewing**: When web applications or UI components are generated, `dsh-moa` integrates seamlessly with `@goodandready/dsh-live-canvas`, automatically spawning sandboxes for 1-click browser previewing.
 5. **Token-Saving Chat Summarization**: Replaces massive code dumps in chat bubbles with compact file listings and clean architectural summaries.
 6. **One-Shot Session Model Restoration**: Executes cleanly as a one-shot turn modifier, automatically reverting back to the user's primary session model immediately after completion.
+7. **Inherent Cost Tracking & Token Estimation**: Embedded pricing calculator for frontier models estimating token usage and run costs directly in the summary badge without external dependencies.
+8. **Refinement Mode (Incremental Edits)**: Automatically detects existing codebase context to generate precise delta modifications instead of destructive full-file rewrites.
+9. **Fast Mode & Custom Judge Criteria**: Ultra-fast single-model preset for quick tasks and customizable evaluation guidelines for the judge.
+10. **Run History & Win-Rate Leaderboard**: Persistent logging with built-in REST endpoints (`/dsh-moa/history` and `/dsh-moa/leaderboard`).
 
 ---
 
@@ -175,6 +179,16 @@ dsh-moa:
 
 ---
 
+## 📊 REST API & Endpoints
+
+| Endpoint | Method | Description |
+|:---|:---|:---|
+| `/dsh-moa/presets` | `GET` | Returns list of configured MoA presets |
+| `/dsh-moa/history?limit=20&offset=0` | `GET` | Returns recent MoA runs with candidates, winner, cost, and tokens |
+| `/dsh-moa/leaderboard` | `GET` | Computes model win-rate leaderboard and average execution costs |
+
+---
+
 ## 🧪 Testing
 
 Run the automated test suite:

package/docs/README.ru.md

@@ -37,6 +37,10 @@
 4. **Мгновенный предпросмотр через Live Canvas**: При создании веб-приложений и UI-компонентов `dsh-moa` бесшовно интегрируется с `@goodandready/dsh-live-canvas`, автоматически инициализируя контейнер для предпросмотра в браузере в 1 клик.
 5. **Экономия токенов в чате**: Вместо вывода огромных листингов кода в чат формируется компактный отчет с перечнем созданных файлов и архитектурным резюме.
 6. **Одноразовый модификатор сессии**: Команда выполняется в рамках одного такта и автоматически возвращает исходную модель сессии пользователя сразу после завершения.
+7. **Встроенный подсчет стоимости и токенов**: Собственный калькулятор стоимости моделей оценивает расход токенов и цену запуска напрямую в бейдже без внешних плагинов.
+8. **Режим доработки (Refinement Mode)**: Автоматически считывает контекст существующего проекта и генерирует точечные дельта-правки без перетирания всей кодовой базы.
+9. **Быстрый режим (Fast Mode) и критерии судьи**: Режим для одиночных быстрых задач без судьи и гибкая настройка фокуса оценки (безопасность, производительность, минимализм).
+10. **История запусков и лидерборд моделей**: Персистентное логирование запусков и REST-эндпоинты (`/dsh-moa/history` и `/dsh-moa/leaderboard`).
 
 ---
 
@@ -175,6 +179,16 @@ dsh-moa:
 
 ---
 
+## 📊 REST API эндпоинты
+
+| Эндпоинт | Метод | Описание |
+|:---|:---|:---|
+| `/dsh-moa/presets` | `GET` | Возвращает список настроенных пресетов MoA |
+| `/dsh-moa/history?limit=20&offset=0` | `GET` | История запусков с кандидатами, победителем, токенами и ценой |
+| `/dsh-moa/leaderboard` | `GET` | Лидерборд побед моделей и средняя стоимость генерации |
+
+---
+
 ## 🧪 Тестирование
 
 Запуск автоматического набора тестов:

package/lib/client.js

@@ -6,6 +6,7 @@ window.__ModuleLoader__.load({
     let React = require('react')
 
     const NS = 'dsh-moa'
+    let localeRegistered = false
 
     let ChevronIcon = null
     try {
@@ -621,6 +622,30 @@ window.__ModuleLoader__.load({
               })
             ),
             React.createElement('div', { className: 'moa-hint' }, t('aggTempHint') || 'Рекомендуется 0.2 – 0.4 для точной проверки и синтеза без галлюцинаций.')
+          ),
+          React.createElement(
+            'div',
+            { className: 'moa-field', style: { marginTop: 4, borderTop: '1px solid var(--dsw-alias-border-l2)', paddingTop: 8 } },
+            React.createElement('label', { className: 'moa-label', style: { fontSize: 12 } }, t('judgeCriteriaLabel') || 'Критерии оценки судьи (опционально):'),
+            React.createElement('textarea', {
+              className: 'moa-textarea',
+              rows: 2,
+              style: {
+                width: '100%',
+                boxSizing: 'border-box',
+                padding: '8px 10px',
+                borderRadius: 8,
+                border: '1px solid var(--dsw-alias-border-l2)',
+                background: 'var(--dsw-alias-bg-layer-3)',
+                color: 'var(--dsw-alias-label-primary)',
+                fontSize: 13,
+                resize: 'vertical',
+                outline: 'none',
+              },
+              placeholder: t('judgeCriteriaPlaceholder') || 'например: Приоритет — производительность, чистый код, минимум зависимостей...',
+              value: judgeCriteria,
+              onChange: (e) => setJudgeCriteria(e.target.value),
+            })
           )
         ),
         /* Proposers / Advisors Box with Temperature Slider */
@@ -911,9 +936,14 @@ window.__ModuleLoader__.load({
       )
     }
 
-    exports.inject = ['slots', 'locale', 'sessions', 'inputTriggers']
+    exports.inject = ['slots', 'locale', 'inputTriggers']
     exports.apply = function apply(ctx) {
-      ctx.locale.register(NS, {
+      const registerLocale = () => {
+        if (localeRegistered) return
+        try {
+          if (ctx.locale && typeof ctx.locale.register === 'function') {
+            localeRegistered = true
+            return ctx.locale.register(NS, {
         en: {
           title: 'Mixture of Agents (MoA)',
           description: 'Run tasks with multiple parallel proposer models and a synthesizing judge model',
@@ -954,6 +984,8 @@ window.__ModuleLoader__.load({
           noModelsMatch: 'No models found matching query.',
           useCustom: 'Use custom',
           selectModel: 'Select model...',
+          judgeCriteriaLabel: 'Custom Judge Evaluation Criteria (optional):',
+          judgeCriteriaPlaceholder: 'e.g. Priority: performance, zero external dependencies, robust edge-case handling...',
         },
         ru: {
           title: 'Mixture of Agents (MoA)',
@@ -995,10 +1027,30 @@ window.__ModuleLoader__.load({
           noModelsMatch: 'Модели не найдены.',
           useCustom: 'Использовать кастомное',
           selectModel: 'Выберите модель...',
+          judgeCriteriaLabel: 'Критерии оценки судьи (опционально):',
+          judgeCriteriaPlaceholder: 'например: Приоритет — производительность, минимум зависимостей, обработка граничных случаев...',
         },
       })
+          }
+        } catch (err) {
+          console.warn('[dsh-moa] locale registration error:', err)
+        }
+      }
+
+      if (typeof ctx.effect === 'function') {
+        ctx.effect(() => registerLocale(), 'dsh-moa: dictionaries')
+      } else {
+        registerLocale()
+      }
 
-      const t = ctx.locale.bind(NS)
+      const t = (key) => {
+        try {
+          if (ctx.locale && typeof ctx.locale.bind === 'function') {
+            return ctx.locale.bind(NS)(key) || key
+          }
+        } catch {}
+        return key
+      }
 
       if (ctx.slots) {
         const registerDirectSection = () => {

package/lib/file-workspace.js

@@ -85,6 +85,89 @@ function sanitizePath(filePath) {
   return normalized.replace(/^[\\\/]+/, '')
 }
 
+/**
+ * Collects readable text files from current workspace to feed as context for candidate models (#6).
+ * Ignores node_modules, .git, .moa, binary files, large bundles.
+ */
+export async function collectProjectContext(baseDir, maxCharBudget = 16000) {
+  if (!baseDir) return { files: [], totalChars: 0 }
+
+  const ignoreDirs = new Set(['node_modules', '.git', '.moa', '.worktrees', 'dist', 'build', '.dsh'])
+  const codeExts = new Set([
+    '.html', '.htm', '.css', '.js', '.mjs', '.cjs', '.jsx', '.ts', '.tsx',
+    '.vue', '.svelte', '.json', '.py', '.md', '.txt', '.sh', '.yaml', '.yml',
+  ])
+
+  const collected = []
+  let totalChars = 0
+
+  async function scan(currentDir, relPrefix = '') {
+    if (totalChars >= maxCharBudget) return
+    try {
+      const list = await fs.readdir(currentDir, { withFileTypes: true })
+      for (const item of list) {
+        if (totalChars >= maxCharBudget) break
+        if (item.name.startsWith('.') && item.name !== '.env') continue
+
+        const rel = relPrefix ? `${relPrefix}/${item.name}` : item.name
+        const full = path.join(currentDir, item.name)
+
+        if (item.isDirectory()) {
+          if (!ignoreDirs.has(item.name)) {
+            await scan(full, rel)
+          }
+        } else if (item.isFile()) {
+          const ext = path.extname(item.name).toLowerCase()
+          if (codeExts.has(ext)) {
+            try {
+              const stat = await fs.stat(full)
+              if (stat.size > 100000) continue // skip files > 100kb
+              const content = await fs.readFile(full, 'utf8')
+              const budgetRemaining = maxCharBudget - totalChars
+              const truncated = content.length > budgetRemaining ? content.slice(0, budgetRemaining) + '\n... [truncated]' : content
+              totalChars += truncated.length
+              collected.push({ relativePath: rel, content: truncated, size: stat.size })
+            } catch {}
+          }
+        }
+      }
+    } catch {}
+  }
+
+  await scan(baseDir)
+  return { files: collected, totalChars }
+}
+
+/**
+ * Checks if a task prompt is an iterative refinement / modification
+ * of an existing project rather than a greenfield project creation (#4).
+ */
+export function isRefinementTask(userPrompt = '', projectFiles = []) {
+  if (!Array.isArray(projectFiles) || projectFiles.length === 0) return false
+  const p = userPrompt.toLowerCase().trim()
+  if (!p) return false
+
+  // Greenfield creation phrases - ALWAYS fresh task
+  const freshPhrases = [
+    'с нуля', 'from scratch', 'новый проект', 'новое приложение', 'new project', 'new app',
+    'создай проект', 'создай приложение', 'создай игру', 'создай сервис',
+    'сделай проект', 'сделай приложение', 'сделай игру',
+    'create a new', 'build a new', 'generate a new',
+  ]
+  if (freshPhrases.some((phrase) => p.includes(phrase))) {
+    return false
+  }
+
+  // Modification keywords
+  const modKeywords = [
+    'добавь', 'измени', 'поменяй', 'исправь', 'обнови', 'переделай', 'доработай', 'удали',
+    'add', 'change', 'update', 'fix', 'modify', 'refactor', 'improve', 'enhance', 'adjust',
+    'style', 'css', 'темн', 'светл', 'color', 'цвет', 'кнопк', 'theme', 'bug', 'баг',
+  ]
+
+  return modKeywords.some((kw) => p.includes(kw))
+}
+
 /**
  * Writes candidate files into an isolated candidate workspace:
  * <baseDir>/.moa/candidate-<index>/
@@ -114,7 +197,7 @@ export async function writeCandidateWorkspace(baseDir, candidateIndex, files = [
  * Promotes winning candidate files to baseDir root,
  * then cleans up the temporary .moa folder.
  */
-export async function promoteCandidateWorkspace(baseDir, winningIndex) {
+export async function promoteCandidateWorkspace(baseDir, winningIndex, options = { keepMoa: false }) {
   if (!baseDir || winningIndex === null || winningIndex === undefined) return []
 
   const winningDir = path.join(baseDir, '.moa', `candidate-${winningIndex}`)
@@ -123,7 +206,9 @@ export async function promoteCandidateWorkspace(baseDir, winningIndex) {
   try {
     const entries = await getFilesRecursively(winningDir)
     if (entries.length === 0) {
-      try { await fs.rm(moaRoot, { recursive: true, force: true }) } catch {}
+      if (!options.keepMoa) {
+        try { await fs.rm(moaRoot, { recursive: true, force: true }) } catch {}
+      }
       return []
     }
 
@@ -137,12 +222,16 @@ export async function promoteCandidateWorkspace(baseDir, winningIndex) {
       promoted.push(rel)
     }
 
-    // Clean up temporary .moa folder
-    await fs.rm(moaRoot, { recursive: true, force: true })
+    if (!options.keepMoa) {
+      // Clean up temporary .moa folder
+      await fs.rm(moaRoot, { recursive: true, force: true })
+    }
 
     return promoted
   } catch (err) {
-    try { await fs.rm(moaRoot, { recursive: true, force: true }) } catch {}
+    if (!options.keepMoa) {
+      try { await fs.rm(moaRoot, { recursive: true, force: true }) } catch {}
+    }
     return []
   }
 }

package/lib/history.js

@@ -0,0 +1,191 @@
+/**
+ * MoA Run History & Analytics Module
+ * Records every MoA pipeline execution, tracks model win rates and costs,
+ * and provides history search and run replay capabilities.
+ */
+
+import fs from 'node:fs'
+import path from 'node:path'
+import os from 'node:os'
+import crypto from 'node:crypto'
+
+const DEFAULT_HISTORY_DIR = path.join(os.homedir(), '.dsh')
+const DEFAULT_HISTORY_FILE = path.join(DEFAULT_HISTORY_DIR, 'moa-history.jsonl')
+
+export function getHistoryFilePath(customDir) {
+  if (customDir) {
+    return path.join(customDir, '.moa-history.jsonl')
+  }
+  return DEFAULT_HISTORY_FILE
+}
+
+/**
+ * Appends a completed MoA run record to the history file.
+ */
+export function recordMoaRun(record, filePath = DEFAULT_HISTORY_FILE) {
+  try {
+    const dir = path.dirname(filePath)
+    if (!fs.existsSync(dir)) {
+      fs.mkdirSync(dir, { recursive: true })
+    }
+
+    const entry = {
+      id: record.id || crypto.randomUUID(),
+      timestamp: record.timestamp || new Date().toISOString(),
+      prompt: record.prompt || '',
+      preset: record.preset || 'default',
+      isRefinement: Boolean(record.isRefinement),
+      candidates: Array.isArray(record.candidates)
+        ? record.candidates.map((c) => ({
+            provider: c.provider || '',
+            model: c.model || '',
+            filesCount: Array.isArray(c.files) ? c.files.length : (c.filesCount || 0),
+            usage: c.usage || { inputTokens: 0, outputTokens: 0 },
+            costUsd: Number(c.costUsd || 0),
+          }))
+        : [],
+      aggregator: record.aggregator
+        ? {
+            provider: record.aggregator.provider || '',
+            model: record.aggregator.model || '',
+            usage: record.aggregator.usage || { inputTokens: 0, outputTokens: 0 },
+            costUsd: Number(record.aggregator.costUsd || 0),
+          }
+        : null,
+      winnerIndex: typeof record.winnerIndex === 'number' ? record.winnerIndex : -1,
+      winnerModel: record.winnerModel || '',
+      promotedFiles: Array.isArray(record.promotedFiles) ? record.promotedFiles : [],
+      totalTokens: Number(record.totalTokens || 0),
+      totalCostUsd: Number(record.totalCostUsd || 0),
+      durationMs: Number(record.durationMs || 0),
+    }
+
+    const line = JSON.stringify(entry) + '\n'
+    fs.appendFileSync(filePath, line, 'utf8')
+    return entry
+  } catch (err) {
+    console.warn('[dsh-moa] Failed to record run to history:', err)
+    return null
+  }
+}
+
+/**
+ * Reads history runs with pagination (latest first).
+ */
+export function getMoaHistory(limit = 20, offset = 0, filePath = DEFAULT_HISTORY_FILE) {
+  try {
+    if (!fs.existsSync(filePath)) {
+      return { total: 0, runs: [] }
+    }
+
+    const raw = fs.readFileSync(filePath, 'utf8')
+    const lines = raw.trim().split('\n').filter(Boolean)
+    const total = lines.length
+
+    const sliced = lines
+      .slice(Math.max(0, total - offset - limit), total - offset)
+      .reverse()
+      .map((line) => {
+        try {
+          return JSON.parse(line)
+        } catch {
+          return null
+        }
+      })
+      .filter(Boolean)
+
+    return { total, runs: sliced }
+  } catch (err) {
+    console.warn('[dsh-moa] Failed to read history:', err)
+    return { total: 0, runs: [] }
+  }
+}
+
+/**
+ * Calculates win-rate leaderboard and model performance statistics across history.
+ */
+export function getMoaLeaderboard(filePath = DEFAULT_HISTORY_FILE) {
+  try {
+    if (!fs.existsSync(filePath)) {
+      return { totalRuns: 0, models: [] }
+    }
+
+    const raw = fs.readFileSync(filePath, 'utf8')
+    const lines = raw.trim().split('\n').filter(Boolean)
+    const totalRuns = lines.length
+
+    const modelStats = new Map()
+
+    for (const line of lines) {
+      try {
+        const item = JSON.parse(line)
+        if (!item) continue
+
+        const winner = item.winnerModel ? String(item.winnerModel).trim() : null
+
+        if (Array.isArray(item.candidates)) {
+          for (const cand of item.candidates) {
+            const key = cand.provider && cand.model ? `${cand.provider}:${cand.model}` : (cand.model || 'unknown')
+            if (!modelStats.has(key)) {
+              modelStats.set(key, {
+                modelKey: key,
+                provider: cand.provider || '',
+                model: cand.model || key,
+                runs: 0,
+                wins: 0,
+                totalTokens: 0,
+                totalCostUsd: 0,
+              })
+            }
+            const st = modelStats.get(key)
+            st.runs += 1
+            const inTok = cand.usage?.inputTokens || 0
+            const outTok = cand.usage?.outputTokens || 0
+            st.totalTokens += inTok + outTok
+            st.totalCostUsd += Number(cand.costUsd || 0)
+
+            if (winner && (winner === key || winner === cand.model || winner.includes(cand.model))) {
+              st.wins += 1
+            }
+          }
+        }
+      } catch {
+        // ignore malformed line
+      }
+    }
+
+    const models = Array.from(modelStats.values())
+      .map((st) => ({
+        ...st,
+        winRate: st.runs > 0 ? Number(((st.wins / st.runs) * 100).toFixed(1)) : 0,
+        avgTokens: st.runs > 0 ? Math.round(st.totalTokens / st.runs) : 0,
+        avgCostUsd: st.runs > 0 ? Number((st.totalCostUsd / st.runs).toFixed(4)) : 0,
+      }))
+      .sort((a, b) => b.wins - a.wins || b.winRate - a.winRate)
+
+    return { totalRuns, models }
+  } catch (err) {
+    console.warn('[dsh-moa] Failed to calculate leaderboard:', err)
+    return { totalRuns: 0, models: [] }
+  }
+}
+
+/**
+ * Retrieves a single history run by ID.
+ */
+export function getMoaRunById(runId, filePath = DEFAULT_HISTORY_FILE) {
+  try {
+    if (!fs.existsSync(filePath)) return null
+    const raw = fs.readFileSync(filePath, 'utf8')
+    const lines = raw.trim().split('\n').filter(Boolean)
+    for (let i = lines.length - 1; i >= 0; i--) {
+      try {
+        const item = JSON.parse(lines[i])
+        if (item && item.id === runId) return item
+      } catch {}
+    }
+    return null
+  } catch {
+    return null
+  }
+}