pan-wizard 3.8.0 → 3.12.0

package/pan-wizard-core/bin/lib/core.cjs

@@ -73,8 +73,75 @@ const MODEL_PROFILES = {
   'pan-distiller':            { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
   // v3.7.0 self-improvement loop — observation-only watchdog
   'pan-experiment-runner':    { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
+  // ADR-0033 bot-army — Release squad
+  'pan-release':              { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
 };
 
+// ─── Effort Profiles (2026-06, adaptive-thinking era) ───────────────────────
+//
+// Per-agent base reasoning effort (low|medium|high|xhigh). `effort` is the
+// primary within-model cost/intelligence dial on current models — it replaced
+// fixed thinking budgets. The base values here mirror the `effort:`
+// frontmatter shipped in agents/*.md (a drift test keeps them in sync).
+//
+// Profile modulation: `budget` steps effort down one level (floor: low) as
+// its cost lever; `quality` and `balanced` keep the base. Per-agent override
+// via config.json → effort_overrides.
+
+const EFFORT_ORDER = ['low', 'medium', 'high', 'xhigh'];
+
+const AGENT_BASE_EFFORT = {
+  // Heavy planning/orchestration/debugging — deepest reasoning
+  'pan-planner':              'xhigh',
+  'pan-conductor':            'xhigh',
+  'pan-debugger':             'xhigh',
+  'pan-plan-checker':         'xhigh',
+  // Execution and verification — thorough but bounded
+  'pan-executor':             'high',
+  'pan-roadmapper':           'high',
+  'pan-verifier':             'high',
+  'pan-integration-checker':  'high',
+  'pan-hardener':             'high',
+  'pan-counterfactual':       'high',
+  'pan-previewer':            'high',
+  'pan-experiment-runner':    'high',
+  'pan-optimizer':            'high',
+  'pan-release':              'high',
+  // Research/synthesis/review — moderate depth
+  'pan-phase-researcher':     'medium',
+  'pan-project-researcher':   'medium',
+  'pan-research-synthesizer': 'medium',
+  'pan-knowledge':            'medium',
+  'pan-distiller':            'medium',
+  'pan-meta-reviewer':        'medium',
+  'pan-reviewer':             'medium',
+  // Mechanical documentation pass — fast and scoped
+  'pan-document_code':        'low',
+};
+
+/**
+ * Resolve the reasoning effort level for an agent under the active profile.
+ * Priority: config.effort_overrides[agent] → base effort modulated by
+ * model_profile (budget steps down one level) → 'medium' for unknown agents.
+ *
+ * @param {string} cwd - Project root directory
+ * @param {string} agentType - e.g. "pan-planner"
+ * @returns {string} One of 'low' | 'medium' | 'high' | 'xhigh'
+ */
+function resolveEffortInternal(cwd, agentType) {
+  const config = loadConfig(cwd);
+  const override = config.effort_overrides?.[agentType];
+  if (typeof override === 'string' && EFFORT_ORDER.includes(override.toLowerCase().trim())) {
+    return override.toLowerCase().trim();
+  }
+  const base = AGENT_BASE_EFFORT[agentType] || 'medium';
+  const profile = config.model_profile || 'balanced';
+  if (profile === 'budget') {
+    return EFFORT_ORDER[Math.max(0, EFFORT_ORDER.indexOf(base) - 1)];
+  }
+  return base;
+}
+
 // ─── Output helpers ───────────────────────────────────────────────────────────
 
 /**
@@ -213,7 +280,13 @@ function loadConfig(cwd) {
       execution: parsed.execution || { default_mode: 'wave_order', rollback_snapshots: true, error_pattern_learning: true },
       focus: parsed.focus || { auto_commit: true },
       model_overrides: parsed.model_overrides || {},
+      effort_overrides: parsed.effort_overrides || {},
       routing: parsed.routing || { strategy: 'static', provider: 'auto' },
+      // ADR-0031: project build/verification commands. null = not configured
+      // (focus-auto --clean-seal then asks or skips rather than guessing).
+      build: parsed.build || null,
+      verification: parsed.verification || null,
+      concurrency: parsed.concurrency || { serial_build: false },
     };
   } catch { // Config missing or malformed — use defaults
     return {
@@ -223,7 +296,11 @@ function loadConfig(cwd) {
       execution: { default_mode: 'wave_order', rollback_snapshots: true, error_pattern_learning: true },
       focus: { auto_commit: true },
       model_overrides: {},
+      effort_overrides: {},
       routing: { strategy: 'static', provider: 'auto' },
+      build: null,
+      verification: null,
+      concurrency: { serial_build: false },
     };
   }
 }
@@ -860,6 +937,9 @@ function scanSourceTodos(cwd) {
 
 module.exports = {
   MODEL_PROFILES,
+  AGENT_BASE_EFFORT,
+  EFFORT_ORDER,
+  resolveEffortInternal,
   PROVIDER_MODELS,
   LEGACY_ALIASES,
   COST_MULTIPLIERS,

package/pan-wizard-core/bin/lib/cost.cjs

@@ -30,7 +30,7 @@
  *   - hit rate: cache_read / (cache_read + input - cache_write) if any cache activity
  *
  * Rate table is approximate — real pricing comes from the provider's API.
- * Rates are US dollars per million tokens, indicative as of 2026-04. Users
+ * Rates are US dollars per million tokens, indicative as of 2026-06. Users
  * can override with `.planning/config.json` → `cost.rates`.
  */
 
@@ -48,21 +48,32 @@ const TOKENS_FILE = 'tokens.jsonl';
  * Override per-model in config.json → cost.rates.
  */
 const DEFAULT_RATES = {
-  // Anthropic
-  'claude-opus-4-7':    { input: 15.0, output: 75.0, cache_read: 1.5,  cache_write: 18.75 },
-  'claude-opus-4-6':    { input: 15.0, output: 75.0, cache_read: 1.5,  cache_write: 18.75 },
+  // Anthropic — verified against platform pricing 2026-06. Opus 4.6+ is $5/$25
+  // (the old $15/$75 Opus pricing ended with the 4.5 generation). Cache rates
+  // follow Anthropic's convention: read ≈ 0.1× input, write ≈ 1.25× input.
+  'claude-fable-5':     { input: 10.0, output: 50.0, cache_read: 1.0,  cache_write: 12.5 },
+  'claude-opus-4-8':    { input: 5.0,  output: 25.0, cache_read: 0.5,  cache_write: 6.25 },
+  'claude-opus-4-7':    { input: 5.0,  output: 25.0, cache_read: 0.5,  cache_write: 6.25 },
+  'claude-opus-4-6':    { input: 5.0,  output: 25.0, cache_read: 0.5,  cache_write: 6.25 },
   'claude-sonnet-4-6':  { input: 3.0,  output: 15.0, cache_read: 0.3,  cache_write: 3.75 },
   'claude-haiku-4-5':   { input: 1.0,  output: 5.0,  cache_read: 0.1,  cache_write: 1.25 },
 
+  // OpenAI — verified against published pricing 2026-06 ($5/$30 standard tier).
+  // Prompt caching is a 90% input discount with no separate write charge, so
+  // cache_write bills at the plain input rate.
+  'gpt-5.5':            { input: 5.0,  output: 30.0, cache_read: 0.5,  cache_write: 5.0 },
+
   // Google Gemini — published rates (per million tokens, approximate; users can override via config.json → cost.rates).
-  // 2.5 tier uses the <=200K-context tier; long-context calls may be billed at ~2x. Cache rates are Google's context-cache pricing (~25% of input rate).
+  // Pro tiers use the <=200K-context tier; long-context calls may be billed at ~2x. Cache rates are Google's context-cache pricing (~25% of input rate).
+  // (gemini-1.5-pro removed 2026-06: retired model; records for it fall back to tier rates.)
+  'gemini-3.1-pro':         { input: 2.00, output: 12.0, cache_read: 0.50,   cache_write: 2.00 },
+  'gemini-3.1-pro-preview': { input: 2.00, output: 12.0, cache_read: 0.50,   cache_write: 2.00 },
   'gemini-2.5-pro':         { input: 1.25, output: 10.0, cache_read: 0.3125, cache_write: 1.25 },
   'gemini-2.5-flash':       { input: 0.30, output: 2.50, cache_read: 0.075,  cache_write: 0.30 },
   'gemini-2.5-flash-lite':  { input: 0.10, output: 0.40, cache_read: 0.025,  cache_write: 0.10 },
-  'gemini-1.5-pro':         { input: 1.25, output: 5.00, cache_read: 0.3125, cache_write: 1.25 },
 
-  // Tier fallbacks when model id is unknown
-  'reasoning': { input: 15.0, output: 75.0, cache_read: 1.5,  cache_write: 18.75 },
+  // Tier fallbacks when model id is unknown (reasoning tracks current Opus pricing)
+  'reasoning': { input: 5.0,  output: 25.0, cache_read: 0.5,  cache_write: 6.25 },
   'mid':       { input: 3.0,  output: 15.0, cache_read: 0.3,  cache_write: 3.75 },
   'fast':      { input: 1.0,  output: 5.0,  cache_read: 0.1,  cache_write: 1.25 },
 };
@@ -81,6 +92,15 @@ function resolveRate(model, tier, configRates) {
     if (tier && configRates[tier]) return configRates[tier];
   }
   if (model && DEFAULT_RATES[model]) return DEFAULT_RATES[model];
+  // Transcript/hook-captured ids are versioned ("claude-opus-4-8-20260301",
+  // "claude-fable-5[1m]") while the table uses family keys — prefix-match,
+  // longest key first so the most specific family wins.
+  if (model) {
+    const families = Object.keys(DEFAULT_RATES)
+      .filter(k => model.startsWith(k))
+      .sort((a, b) => b.length - a.length);
+    if (families.length > 0) return DEFAULT_RATES[families[0]];
+  }
   if (tier && DEFAULT_RATES[tier]) return DEFAULT_RATES[tier];
   return null;
 }
@@ -342,6 +362,37 @@ function cmdCostClear(cwd, raw) {
   }
 }
 
+// ─── Rate-table staleness ───────────────────────────────────────────────────
+
+// Date DEFAULT_RATES was last verified against published provider pricing.
+// Bump this whenever the table is re-verified; `models check` flags the table
+// once it is older than RATES_STALE_AFTER_DAYS (provider prices move faster
+// than PAN releases do).
+const RATES_VERIFIED_AT = '2026-06-10';
+const RATES_STALE_AFTER_DAYS = 180;
+const RATE_TIERS = ['reasoning', 'mid', 'fast'];
+
+function checkRatesStaleness(now = new Date()) {
+  const verified = new Date(RATES_VERIFIED_AT + 'T00:00:00Z');
+  const ageDays = Math.floor((now.getTime() - verified.getTime()) / 86400000);
+  return {
+    rates_verified_at: RATES_VERIFIED_AT,
+    age_days: ageDays,
+    stale_after_days: RATES_STALE_AFTER_DAYS,
+    stale: ageDays > RATES_STALE_AFTER_DAYS,
+    models: Object.keys(DEFAULT_RATES).filter(k => !RATE_TIERS.includes(k)),
+    tiers: RATE_TIERS,
+  };
+}
+
+function cmdModelsCheck(raw) {
+  const result = checkRatesStaleness();
+  const human = result.stale
+    ? `Rate table verified ${result.rates_verified_at} (${result.age_days} days ago) — STALE: re-verify provider pricing and bump RATES_VERIFIED_AT in cost.cjs`
+    : `Rate table verified ${result.rates_verified_at} (${result.age_days} days ago) — OK`;
+  output(result, raw, human);
+}
+
 module.exports = {
   computeCost,
   appendRecord,
@@ -350,10 +401,13 @@ module.exports = {
   renderTable,
   renderChart,
   resolveRate,
+  checkRatesStaleness,
   cmdCostReport,
   cmdCostAppend,
   cmdCostClear,
+  cmdModelsCheck,
   METRICS_DIR,
   TOKENS_FILE,
   DEFAULT_RATES,
+  RATES_VERIFIED_AT,
 };

package/pan-wizard-core/bin/lib/focus.cjs

@@ -14,7 +14,7 @@ const {
   FOCUS_MODES, FOCUS_TIERS, FOCUS_DIR,
   BUDGET_LIMIT_BUGFIX, BUDGET_LIMIT_FULL, STABILITY_RATIO, FEATURE_RATIO,
   DIMINISHING_RETURNS_THRESHOLD,
-  AUTO_RUN_FILE, FOCUS_CATEGORIES, CATEGORY_PRIORITY_RANGE, CATEGORY_DEFAULTS,
+  AUTO_RUN_FILE, FOCUS_CATEGORIES, FOCUS_SOURCES, CATEGORY_PRIORITY_RANGE, CATEGORY_DEFAULTS,
   DEFAULT_MAX_CYCLES, DEFAULT_TOTAL_BUDGET,
   BUDGET_MIN, BUDGET_MAX, MAX_CYCLES_MIN, MAX_CYCLES_MAX, TOTAL_BUDGET_MIN, TOTAL_BUDGET_MAX,
   AUTORUN_STATUSES, DOC_SYNC_FILES, COMMAND_RENAME_MAP,
@@ -829,6 +829,14 @@ function focusAutoInit(cwd, raw, getVal, hasFlag) {
     return error(`Category must be one of: ${FOCUS_CATEGORIES.join(', ')}`);
   }
 
+  // ADR-0031: work source — 'scan' (category code-scan, default) or 'backlog'
+  // (rank actionable roadmap.md / requirements.md items). Category applies to
+  // scan mode; backlog mode ranks the whole actionable backlog.
+  const source = getVal('--source', 'scan');
+  if (!FOCUS_SOURCES.includes(source)) {
+    return error(`Source must be one of: ${FOCUS_SOURCES.join(', ')}`);
+  }
+
   const existing = readAutoRun(cwd);
   if (existing && (existing.status === AUTORUN_STATUSES.IN_PROGRESS || existing.status === AUTORUN_STATUSES.INITIALIZED)) {
     return error('Auto-run already in progress. Use --stop to end it, or --continue to resume.');
@@ -848,8 +856,12 @@ function focusAutoInit(cwd, raw, getVal, hasFlag) {
   const runData = {
     run_id: generateRunId(cwd),
     status: AUTORUN_STATUSES.INITIALIZED,
+    source: source,
     category: category,
     mode: mode,
+    parallel_research: hasFlag('--parallel-research'),
+    parallel_verify: hasFlag('--parallel-verify'),
+    clean_seal: hasFlag('--clean-seal'),
     budget_per_cycle: budget,
     max_cycles: maxCycles,
     total_budget: totalBudget,

package/pan-wizard-core/bin/lib/git.cjs

@@ -272,7 +272,12 @@ function cmdGitTag(cwd, sub, opts, raw) {
   }
   if (sub === 'create') {
     if (!name) { error('--name required for tag create'); }
-    const args = message ? ['tag', '-m', message, name] : ['tag', name];
+    // tag.gpgsign=true in user config would force signing (and fail outright
+    // for lightweight tags) in non-interactive runs — PAN tags are automation
+    // markers, so signing is explicitly disabled.
+    const args = message
+      ? ['-c', 'tag.gpgsign=false', 'tag', '-m', message, name]
+      : ['-c', 'tag.gpgsign=false', 'tag', name];
     const r = execGit(cwd, args);
     if (r.exitCode !== 0) {
       output({ created: false, tag: name, detail: r.stderr }, raw, 'tag create failed');