dual-brain 4.6.0 → 4.7.1

package/orchestrator.json

@@ -3,87 +3,66 @@
     "claude": {
       "plan": "$100",
       "models": {
-        "opus": {
-          "tier": "think",
-          "input_per_mtok": 5.0,
-          "output_per_mtok": 25.0,
-          "context_window": 1000000,
-          "max_output": 128000,
-          "model_id": "claude-opus-4-6",
-          "strengths": ["agentic coding", "complex reasoning", "tool use", "error recovery", "architecture decisions"],
-          "weaknesses": ["higher latency", "tokenizer tax on 4.7 (12-35% more tokens)"],
-          "best_for": "architecture, security review, complex debugging, multi-step planning",
-          "avoid_for": "simple file reads, grep, formatting — wasteful at this tier"
-        },
-        "sonnet": {
-          "tier": "execute",
-          "input_per_mtok": 3.0,
-          "output_per_mtok": 15.0,
-          "context_window": 1000000,
-          "max_output": 64000,
-          "model_id": "claude-sonnet-4-6",
-          "strengths": ["best speed/intelligence ratio", "precise minimal diffs", "1M context", "extended thinking"],
-          "weaknesses": ["less reliable on complex multi-step reasoning than Opus"],
-          "best_for": "implementation, refactoring, test writing, code edits, git operations",
-          "avoid_for": "architecture decisions, security audits — upgrade to think tier"
-        },
-        "haiku": {
-          "tier": "search",
-          "input_per_mtok": 1.0,
-          "output_per_mtok": 5.0,
-          "context_window": 200000,
-          "max_output": 64000,
-          "model_id": "claude-haiku-4-5-20251001",
-          "strengths": ["fastest latency", "cheapest", "good enough for read-only tasks"],
-          "weaknesses": ["200k context (vs 1M for others)", "weaker reasoning", "older knowledge cutoff"],
-          "best_for": "file lookups, grep, explore, read-only research, listing files",
-          "avoid_for": "any task requiring edits, reasoning, or judgment"
-        }
+        "opus":   { "tier": "think",   "input_per_mtok": 5.0,   "output_per_mtok": 25.0,  "context_window": 1000000, "max_output": 128000 },
+        "sonnet": { "tier": "execute", "input_per_mtok": 3.0,   "output_per_mtok": 15.0,  "context_window": 1000000, "max_output": 64000 },
+        "haiku":  { "tier": "search",  "input_per_mtok": 1.0,   "output_per_mtok": 5.0,   "context_window": 200000,  "max_output": 64000 }
       }
     },
     "openai": {
       "plan": "$100",
       "models": {
-        "gpt-5.5": {
-          "tier": "think",
-          "input_per_mtok": 5.0,
-          "output_per_mtok": 30.0,
-          "context_window": 1000000,
-          "max_output": 128000,
-          "model_id": "gpt-5.5",
-          "strengths": ["complex reasoning", "fast interactive responses", "strong code review", "independent perspective from Claude"],
-          "weaknesses": ["different failure modes than Claude — feature not bug for dual-brain"],
-          "best_for": "independent code review (dual-brain), second opinions on architecture",
-          "codex_compatible": true
-        },
-        "gpt-5.4": {
-          "tier": "execute",
-          "input_per_mtok": 2.5,
-          "output_per_mtok": 15.0,
-          "context_window": 1000000,
-          "max_output": 128000,
-          "model_id": "gpt-5.4",
-          "strengths": ["good speed/cost ratio", "1M context", "strong instruction following"],
-          "weaknesses": ["less capable reasoning than gpt-5.5"],
-          "best_for": "implementation and execution tasks via Codex",
-          "codex_compatible": true
-        },
-        "gpt-4.1-mini": {
-          "tier": "search",
-          "input_per_mtok": 0.40,
-          "output_per_mtok": 1.60,
-          "context_window": 1047576,
-          "max_output": 32768,
-          "model_id": "gpt-4.1-mini",
-          "strengths": ["extremely cheap", "good instruction following", "1M context"],
-          "weaknesses": ["32k max output", "no reasoning chain"],
-          "best_for": "cheap search/lookup tasks via Codex",
-          "codex_compatible": true
-        }
+        "gpt-5.5":      { "tier": "think",   "input_per_mtok": 5.0,   "output_per_mtok": 30.0,  "context_window": 1000000, "max_output": 128000 },
+        "gpt-5.4":      { "tier": "execute", "input_per_mtok": 2.5,   "output_per_mtok": 15.0,  "context_window": 1000000, "max_output": 128000 },
+        "gpt-4.1-mini": { "tier": "search",  "input_per_mtok": 0.40,  "output_per_mtok": 1.60,  "context_window": 1047576, "max_output": 32768 }
       }
     }
   },
 
+  "model_intelligence": {
+    "opus": {
+      "model_id": "claude-opus-4-6",
+      "strengths": ["agentic coding", "complex reasoning", "tool use", "error recovery", "architecture decisions"],
+      "weaknesses": ["higher latency", "tokenizer tax on 4.7 (12-35% more tokens)"],
+      "best_for": "architecture, security review, complex debugging, multi-step planning",
+      "avoid_for": "simple file reads, grep, formatting — wasteful at this tier"
+    },
+    "sonnet": {
+      "model_id": "claude-sonnet-4-6",
+      "strengths": ["best speed/intelligence ratio", "precise minimal diffs", "1M context", "extended thinking"],
+      "weaknesses": ["less reliable on complex multi-step reasoning than Opus"],
+      "best_for": "implementation, refactoring, test writing, code edits, git operations",
+      "avoid_for": "architecture decisions, security audits — upgrade to think tier"
+    },
+    "haiku": {
+      "model_id": "claude-haiku-4-5-20251001",
+      "strengths": ["fastest latency", "cheapest", "good enough for read-only tasks"],
+      "weaknesses": ["200k context (vs 1M for others)", "weaker reasoning", "older knowledge cutoff"],
+      "best_for": "file lookups, grep, explore, read-only research, listing files",
+      "avoid_for": "any task requiring edits, reasoning, or judgment"
+    },
+    "gpt-5.5": {
+      "model_id": "gpt-5.5",
+      "strengths": ["complex reasoning", "fast interactive responses", "strong code review", "independent perspective from Claude"],
+      "weaknesses": ["different failure modes than Claude — feature not bug for dual-brain"],
+      "best_for": "independent code review (dual-brain), second opinions on architecture",
+      "codex_compatible": true
+    },
+    "gpt-5.4": {
+      "model_id": "gpt-5.4",
+      "strengths": ["good speed/cost ratio", "1M context", "strong instruction following"],
+      "weaknesses": ["less capable reasoning than gpt-5.5"],
+      "best_for": "implementation and execution tasks via Codex",
+      "codex_compatible": true
+    },
+    "gpt-4.1-mini": {
+      "model_id": "gpt-4.1-mini",
+      "strengths": ["extremely cheap", "good instruction following", "1M context"],
+      "weaknesses": ["32k max output", "no reasoning chain"],
+      "best_for": "cheap search/lookup tasks via Codex",
+      "codex_compatible": true
+    }
+  },
+
   "tiers": {
     "search": {
       "description": "Read-only lookups, exploration, grep, find, file reads",
@@ -121,6 +100,46 @@
     "model_routing_note": "Claude Code model: param may be silently ignored (issue #43869). Set CLAUDE_CODE_SUBAGENT_MODEL env var as fallback if subagents all run on parent model."
   },
 
+  "codex_skills": {
+    "review": {
+      "description": "Independent GPT code review via ChatGPT subscription",
+      "command": "node .claude/hooks/dual-brain-review.mjs",
+      "requires": "codex_auth"
+    },
+    "quality_gate": {
+      "description": "Config-driven quality gate with review artifacts",
+      "command": "node .claude/hooks/quality-gate.mjs"
+    },
+    "cost_report": {
+      "description": "Session activity and cost estimate by model tier",
+      "command": "node .claude/hooks/cost-report.mjs"
+    },
+    "test": {
+      "description": "Self-test harness for all orchestrator hooks",
+      "command": "node .claude/hooks/test-orchestrator.mjs"
+    },
+    "health_check": {
+      "description": "Verify all hooks are wired and system is healthy",
+      "command": "node .claude/hooks/health-check.mjs"
+    },
+    "session_report": {
+      "description": "Comprehensive session-end summary: activity by tier, routing compliance, quality gate status, data quality, drift warnings",
+      "command": "node .claude/hooks/session-report.mjs"
+    },
+    "setup_wizard": {
+      "description": "Interactive setup wizard — configures Claude Code hooks, dual-provider routing (Claude + OpenAI/Codex), subscription tiers, and cost-tracking settings",
+      "command": "node .claude/hooks/setup-wizard.mjs"
+    },
+    "gpt_dispatch": {
+      "description": "Dispatch execution tasks to GPT via Codex CLI",
+      "command": "node .claude/hooks/gpt-work-dispatcher.mjs"
+    },
+    "budget_balance": {
+      "description": "Show provider balance status and routing recommendations",
+      "command": "node .claude/hooks/budget-balancer.mjs"
+    }
+  },
+
   "pricing_verified": "2026-05-13",
 
   "budgets": {
@@ -134,7 +153,11 @@
   "routing": {
     "strategy": "hybrid-specialized-balanced",
     "codex_startup_penalty_ms": 20000,
-    "min_codex_task_ms": 180000
+    "min_codex_task_ms": 180000,
+    "dual_thinking_permission_threshold": {
+      "estimated_tokens": 50000,
+      "agent_count": 2
+    }
   },
 
   "providers": {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "dual-brain",
-  "version": "4.6.0",
-  "description": "Data-Tools Dual-Brain — dual-provider orchestration extension for data-tools/replit-tools. Tiered routing, budget balancing, and GPT dual-brain review across Claude + OpenAI subscriptions",
+  "version": "4.7.1",
+  "description": "Dual-provider orchestration for Claude Code — tiered routing, budget balancing, and GPT dual-brain review across Claude + OpenAI subscriptions",
   "type": "module",
   "bin": {
     "dual-brain": "./install.mjs"
@@ -25,11 +25,6 @@
   "scripts": {
     "test": "node hooks/test-orchestrator.mjs"
   },
-  "dataTools": {
-    "extension": true,
-    "requires": "replit-tools",
-    "credit": "Built on data-tools by Steve Moraco"
-  },
   "engines": {
     "node": ">=20.0.0"
   },

package/hooks/agent-chains.mjs

@@ -1,369 +0,0 @@
-#!/usr/bin/env node
-/**
- * agent-chains.mjs — Opinionated multi-step agent workflows.
- *
- * Three built-in chains that map to how developers actually work:
- *
- *   explore-then-fix   — Understand the problem, then fix it
- *   review-and-test    — Review the code, then write tests for weak spots
- *   audit-and-plan     — Audit the architecture, then create an execution plan
- *
- * Each chain:
- *   1. Prints a banner showing the steps
- *   2. Runs step 1 (capture output)
- *   3. Prints step 1 results
- *   4. Asks for confirmation (or auto-continues with --yes)
- *   5. Runs step 2 with step 1 output as context
- *   6. Prints final results
- *
- * Export: getChain(name), listChains()
- * CLI: node agent-chains.mjs --list
- *      node agent-chains.mjs --run <chain> [flags]
- */
-
-import { spawnSync } from 'child_process';
-import { createInterface } from 'readline';
-import { dirname, resolve } from 'path';
-import { fileURLToPath } from 'url';
-
-const __dirname = dirname(fileURLToPath(import.meta.url));
-const TEMPLATES_SCRIPT = resolve(__dirname, 'agent-templates.mjs');
-
-// ─── Chain Definitions ─────────────────────────────────────────────────────
-
-/**
- * Built-in chains.
- *
- * Each chain has:
- *   name        - unique identifier
- *   description - one-line description
- *   steps       - array of step descriptors
- *     step.label       - human-readable step label
- *     step.template    - agent-templates.mjs template name
- *     step.tier        - search | execute | think
- *     step.model       - haiku | sonnet | opus
- *     step.args        - function(flags, prevOutput) → args object for the template
- *     step.stop_after  - if true, pause here for confirmation before continuing
- *     step.stop_label  - label for the stop-point prompt
- */
-export const CHAINS = {
-  'explore-then-fix': {
-    name: 'explore-then-fix',
-    description: 'Understand the problem, then fix it',
-    steps: [
-      {
-        label: 'Step 1: Explore',
-        template: 'explorer',
-        tier: 'search',
-        model: 'haiku',
-        args: (flags) => ({
-          question: flags.question,
-          scope: flags.scope,
-        }),
-        stop_after: true,
-        stop_label: 'Exploration complete. Review findings above, then confirm the fix.',
-      },
-      {
-        label: 'Step 2: Fix',
-        template: 'bug-hunter',
-        tier: 'execute',
-        model: 'sonnet',
-        args: (flags, prevOutput) => ({
-          question: flags.question,
-          scope: flags.scope,
-          context: prevOutput,
-        }),
-      },
-    ],
-  },
-
-  'review-and-test': {
-    name: 'review-and-test',
-    description: 'Review the code, then write tests for weak spots',
-    steps: [
-      {
-        label: 'Step 1: Security Review',
-        template: 'security-review',
-        tier: 'think',
-        model: 'opus',
-        args: (flags) => ({
-          scope: flags.scope,
-          file: flags.file,
-        }),
-        stop_after: true,
-        stop_label: 'Review complete. Review findings above, then confirm test writing.',
-      },
-      {
-        label: 'Step 2: Write Tests',
-        template: 'test-writer',
-        tier: 'execute',
-        model: 'sonnet',
-        args: (flags, prevOutput) => ({
-          scope: flags.scope,
-          file: flags.file,
-          context: prevOutput,
-        }),
-      },
-    ],
-  },
-
-  'audit-and-plan': {
-    name: 'audit-and-plan',
-    description: 'Audit the architecture, then create an execution plan',
-    steps: [
-      {
-        label: 'Step 1: Architecture Audit',
-        template: 'explorer',
-        tier: 'search',
-        model: 'haiku',
-        args: (flags) => ({
-          question: flags.question,
-          scope: flags.scope,
-        }),
-        stop_after: true,
-        stop_label: 'Audit complete. Review findings above, then confirm planning.',
-      },
-      {
-        label: 'Step 2: Create Plan',
-        template: null, // Think tier — runs via plan-generator concept inline
-        tier: 'think',
-        model: 'opus',
-        args: (flags, prevOutput) => ({
-          question: flags.question,
-          scope: flags.scope,
-          context: prevOutput,
-        }),
-        // Uses a custom prompt builder instead of an agent template
-        custom_prompt: (flags, prevOutput) => {
-          const scopeLine = flags.scope ? `\nScope: ${flags.scope}` : '';
-          return [
-            `You are an architect. Based on the audit findings below, create a detailed execution plan.`,
-            ``,
-            `Question / Goal: ${flags.question || 'Improve the architecture based on audit findings'}`,
-            scopeLine,
-            ``,
-            `Audit findings:`,
-            prevOutput,
-            ``,
-            `Produce a structured plan with:`,
-            `- Decision: the recommended approach`,
-            `- Rationale: why this approach over alternatives`,
-            `- Alternatives considered (and why rejected)`,
-            `- Risks and mitigations`,
-            `- Verification plan: how to confirm success`,
-            `- Task table: dependency-ordered list of concrete tasks with tier/risk`,
-          ].join('\n');
-        },
-      },
-    ],
-  },
-};
-
-// ─── Exports ───────────────────────────────────────────────────────────────
-
-/**
- * Get a chain by name. Returns null if not found.
- */
-export function getChain(name) {
-  return CHAINS[name] || null;
-}
-
-/**
- * List all chains as { name, description, steps }.
- */
-export function listChains() {
-  return Object.values(CHAINS).map(({ name, description, steps }) => ({
-    name,
-    description,
-    steps: steps.map(s => ({
-      label: s.label,
-      template: s.template,
-      tier: s.tier,
-      model: s.model,
-    })),
-  }));
-}
-
-// ─── Chain Execution ───────────────────────────────────────────────────────
-
-function banner(chain) {
-  const width = 60;
-  const line = '─'.repeat(width);
-  console.log(`\n  ${line}`);
-  console.log(`  Chain: ${chain.name} — ${chain.description}`);
-  console.log(`  ${line}`);
-  for (let i = 0; i < chain.steps.length; i++) {
-    const s = chain.steps[i];
-    const model = s.model || (s.tier === 'think' ? 'opus' : s.tier === 'search' ? 'haiku' : 'sonnet');
-    const tmpl = s.template || 'custom';
-    console.log(`    ${i + 1}. ${s.label}  [${s.tier} / ${model} / ${tmpl}]`);
-  }
-  console.log(`  ${line}\n`);
-}
-
-function separator(label) {
-  const width = 60;
-  const line = '─'.repeat(width);
-  console.log(`\n  ${line}`);
-  if (label) console.log(`  ${label}`);
-  console.log(`  ${line}\n`);
-}
-
-function runTemplate(templateName, templateArgs) {
-  const argsList = ['--run', templateName];
-  if (templateArgs.question) argsList.push('--question', templateArgs.question);
-  if (templateArgs.scope)    argsList.push('--scope', templateArgs.scope);
-  if (templateArgs.file)     argsList.push('--file', templateArgs.file);
-  if (templateArgs.context)  argsList.push('--context', templateArgs.context);
-
-  const result = spawnSync(process.execPath, [TEMPLATES_SCRIPT, ...argsList], {
-    encoding: 'utf8',
-    stdio: ['pipe', 'pipe', 'inherit'],
-    timeout: 30_000,
-  });
-
-  return {
-    output: result.stdout || '',
-    status: result.status,
-    stderr: result.stderr || '',
-  };
-}
-
-function runCustomStep(step, flags, prevOutput) {
-  const prompt = step.custom_prompt(flags, prevOutput);
-  const model = step.model || 'opus';
-  const tier = step.tier || 'think';
-
-  console.log(`\n  [${step.label}] tier=${tier} model=${model} custom\n`);
-  console.log('─'.repeat(60));
-  console.log(prompt);
-  console.log('─'.repeat(60));
-
-  return {
-    output: prompt,
-    status: 0,
-  };
-}
-
-async function askConfirmation(stopLabel) {
-  console.log(`\n  ${stopLabel}`);
-  const rl = createInterface({ input: process.stdin, output: process.stdout });
-  return new Promise((resolve) => {
-    rl.question('\n  Continue to step 2? [Y/n] ', (answer) => {
-      rl.close();
-      const ans = answer.trim().toLowerCase();
-      resolve(ans === '' || ans === 'y' || ans === 'yes');
-    });
-  });
-}
-
-async function executeChain(chain, flags) {
-  banner(chain);
-
-  let prevOutput = '';
-
-  for (let i = 0; i < chain.steps.length; i++) {
-    const step = chain.steps[i];
-    const isLast = i === chain.steps.length - 1;
-
-    console.log(`\n  Running ${step.label}...`);
-
-    let result;
-    if (step.template) {
-      const templateArgs = step.args(flags, prevOutput);
-      result = runTemplate(step.template, templateArgs);
-    } else if (step.custom_prompt) {
-      result = runCustomStep(step, flags, prevOutput);
-    } else {
-      console.error(`  Step "${step.label}" has no template or custom_prompt.`);
-      process.exit(1);
-    }
-
-    if (result.status !== 0) {
-      console.error(`\n  Step failed (exit ${result.status}). Chain aborted.`);
-      process.exit(result.status || 1);
-    }
-
-    // Print step output
-    separator(`Results: ${step.label}`);
-    if (result.output.trim()) {
-      console.log(result.output);
-    } else {
-      console.log('  (no output)');
-    }
-
-    prevOutput = result.output;
-
-    // Stop point: ask for confirmation before next step
-    if (step.stop_after && !isLast) {
-      if (flags.yes) {
-        console.log(`\n  [--yes] Auto-continuing to next step...`);
-      } else {
-        const confirmed = await askConfirmation(step.stop_label || `Step ${i + 1} complete.`);
-        if (!confirmed) {
-          console.log('\n  Aborted by user. Findings saved above.\n');
-          process.exit(0);
-        }
-      }
-    }
-  }
-
-  separator('Chain Complete');
-  console.log('  All steps finished. Review results above.\n');
-}
-
-// ─── CLI ───────────────────────────────────────────────────────────────────
-
-function parseArgs(argv) {
-  const flags = { yes: false };
-  for (let i = 0; i < argv.length; i++) {
-    const a = argv[i];
-    if (a === '--list')          flags.list = true;
-    else if (a === '--run')      flags.run = argv[++i];
-    else if (a === '--question') flags.question = argv[++i];
-    else if (a === '--scope')    flags.scope = argv[++i];
-    else if (a === '--file')     flags.file = argv[++i];
-    else if (a === '--yes' || a === '-y') flags.yes = true;
-  }
-  return flags;
-}
-
-if (process.argv[1] && fileURLToPath(import.meta.url) === resolve(process.argv[1])) {
-  const flags = parseArgs(process.argv.slice(2));
-
-  if (flags.list) {
-    console.log('\n  Agent Chains:\n');
-    for (const c of listChains()) {
-      console.log(`    ${c.name.padEnd(22)} ${c.description}`);
-      for (const s of c.steps) {
-        const tmpl = s.template || 'custom';
-        console.log(`      ${s.label.padEnd(30)} [${s.tier}/${s.model}/${tmpl}]`);
-      }
-      console.log('');
-    }
-    process.exit(0);
-  }
-
-  if (flags.run) {
-    const chain = getChain(flags.run);
-    if (!chain) {
-      console.error(`  Unknown chain: ${flags.run}`);
-      console.error(`  Available: ${Object.keys(CHAINS).join(', ')}`);
-      process.exit(1);
-    }
-
-    executeChain(chain, flags).catch((err) => {
-      console.error(`  Chain error: ${err.message}`);
-      process.exit(1);
-    });
-  } else {
-    console.log(`
-  Usage:
-    node agent-chains.mjs --list
-    node agent-chains.mjs --run explore-then-fix --question "what's wrong with auth" [--scope "src/auth"] [--yes]
-    node agent-chains.mjs --run review-and-test [--scope "src/api"] [--file "src/api.ts"] [--yes]
-    node agent-chains.mjs --run audit-and-plan --question "how should we restructure auth" [--scope "src/"] [--yes]
-    `);
-    process.exit(0);
-  }
-}