pan-wizard 3.18.0 → 3.19.0

package/bin/install-lib.cjs

@@ -1016,9 +1016,18 @@ function detectModelCapabilities(modelName) {
   const n = modelName.toLowerCase();
 
   // Anthropic Claude family
-  if (n.includes('fable')) {
+  if (n.includes('fable') || n.includes('mythos')) {
     return { has_1m_ctx: true, has_thinking: true, has_cache: true, tier: 'reasoning' };
   }
+  // Claude 5 family (Opus 5, Sonnet 5) — 1M context, extended thinking, prompt caching.
+  // Without this, `claude-opus-5` falls through to `unknown` and the installer prints
+  // a FALSE "your model lacks 1M context / extended thinking" warning on the flagship.
+  if (n.includes('opus-5')) {
+    return { has_1m_ctx: true, has_thinking: true, has_cache: true, tier: 'reasoning' };
+  }
+  if (n.includes('sonnet-5')) {
+    return { has_1m_ctx: true, has_thinking: true, has_cache: true, tier: 'mid' };
+  }
   if (n.includes('opus-4-8') || n.includes('opus-4.8')
     || n.includes('opus-4-7') || n.includes('opus-4.7')
     || n.includes('opus-4-6') || n.includes('opus-4.6')) {

package/bin/install.js

@@ -2492,7 +2492,7 @@ function finishInstall(settingsPath, settings, statuslineCommand, shouldInstallS
           console.log(`
   ${yellow}ℹ${reset} PAN's multi-agent workflows are tuned for frontier reasoning models. Default model "${modelField}" lacks: ${missing}.
      Features degrade gracefully, but for best results select claude-fable-5 (PAN's recommended flagship — deepest
-     long-horizon reasoning for the bot army) or claude-opus-4-8 (same 1M context at half the cost).`);
+     long-horizon reasoning for the bot army), or an Opus-tier model (claude-opus-5 / claude-opus-4-8) at lower cost.`);
         }
       }
     } catch {

package/commands/pan/army.md

@@ -78,13 +78,14 @@ Every cap the conductor enforces applies to the campaign, scaled up:
 |------|---------|--------|
 | `--source` | `backlog` | Work selection (delegates to focus-auto): `backlog` = ranked roadmap/requirements items; `scan` = category code-scan. |
 | `--max-cycles` | 5 | Mission items landed before stopping. |
-| `--total-budget` | 300 | Cumulative point ceiling. |
+| `--total-budget` | 300 | Cumulative point budget. **Advisory by default** — tracked/surfaced but not a hard stop unless `--enforce-budget` / config `budget.enforce: true`. |
+| `--enforce-budget` | off | Make the point budgets hard stops again (also settable via config `budget.enforce: true`). |
 | `--squads` | all | Restrict to a subset, e.g. `--squads architecture,build,quality`. |
 | `--no-build-worktrees` | off | Build in the main tree instead of branch-per-agent worktrees (small/serial projects). |
 | `--push` | off | Push approved merges to origin (still human-gated). |
 | `--clean-seal` | off | One clean build + full verification after the last item (commands from config). |
 | `--schedule` | off | Arm a self-resuming campaign at this cadence (`hourly`/`daily`/`weekly`/`Nh`/`Nd`) instead of running once — writes the schedule descriptor (ADR-0034). Pair with `--daily-budget`. |
-| `--daily-budget` | 300 | Per-day point ceiling for a scheduled campaign; the day's run stops when reached, resumes next day. |
+| `--daily-budget` | 300 | Per-day point budget for a scheduled campaign. Advisory by default (an indicator of the day's spend); it only pauses the day's run when `budget.enforce`/`enforce_budget` is set. |
 | `--dry-run` | off | Plan + squad delegation preview only; STOP. |
 | `--continue` / `--stop` / `--status` | — | Resume / halt / report from `.planning/orchestration/` + focus-auto state. |
 
@@ -141,14 +142,14 @@ PAN is not a daemon — it cannot wake itself while the session is closed. `--sc
 - **Arm:** `/pan:army "<goal>" --schedule daily --daily-budget 200` writes `.planning/orchestration/schedule.json` (cadence, daily budget, next-due) instead of running once.
 - **The trigger (you wire one):** a host scheduler (Claude Code routines / cron / scheduled-tasks) or a `/loop` runs `pan-tools campaign due` and, when it reports due, invokes `/pan:army --continue`. On next session open, a due campaign is surfaced as a nudge.
 - **Resume (`--continue`):** read the schedule + `.planning/orchestration/` + focus-auto state. If `campaign due` is true and the day's `--daily-budget` isn't spent, run the next mission(s), then `campaign record-run` (advances next-due, accrues the day's spend). If not due or budget-spent, report next-due and STOP.
-- **Bounded spend:** the per-day budget caps each day's run; the per-run `--total-budget` and the conductor caps still bound each cycle. A scheduled campaign runs the backlog down to staged, reviewed, green PRs over days — and still waits for a human at every merge.
+- **Bounded spend:** point budgets (`--total-budget`, `--daily-budget`) are **advisory indicators by default** — they're tracked and surfaced, not hard stops, unless `budget.enforce` / `--enforce-budget` is set. The real bounds are `--max-cycles`, the conductor caps, the abort file, and the human merge gate at every integrate. A scheduled campaign runs the backlog down to staged, reviewed, green PRs over days.
 
 Manage it: `pan-tools campaign status` (active/paused, spent today, next-due), `campaign schedule --pause` / `--resume` / `--disable`.
 
 ---
 
 ## Completion contract
-The campaign is complete when ANY holds: `--max-cycles` reached · `--total-budget` exhausted · backlog empty · abort file present · context < 25% · a mission cannot pass Quality and can't be cleanly reverted (HARD STOP — preserve state, report). Always run `--clean-seal` (unless omitted) after the last item.
+The campaign is complete when ANY holds: `--max-cycles` reached · backlog empty · abort file present · context < 25% · a mission cannot pass Quality and can't be cleanly reverted (HARD STOP — preserve state, report). (Budget exhaustion is advisory by default — a stop condition only when `budget.enforce`/`--enforce-budget` is set.) Always run `--clean-seal` (unless omitted) after the last item.
 
 ## NEVER DO
 - Let Mission Control write code, or let a squad agent spawn further agents (depth cap).

package/commands/pan/focus-auto.md

@@ -33,7 +33,7 @@ This command runs improvement campaigns on the **host project's source code**
 <completion_contract>
 A campaign is complete when ANY stop condition is met:
 1. Max cycles reached (--max-cycles, default 10)
-2. Total budget exhausted (--total-budget, default 200)
+2. Total budget exhausted (--total-budget) — **advisory by default**: tracked and surfaced but does NOT stop the run unless `--enforce-budget` (or config `budget.enforce: true`) is set
 3. Scan returns zero items for the selected category
 4. Context window drops below 25% (CRITICAL threshold)
 5. User sends /pan:focus-auto --stop
@@ -100,7 +100,8 @@ Wait for the user's reply before proceeding. Do not guess or pick a default cate
 | `--mode` | category-dependent | bugfix, balanced, features, full |
 | `--budget` | category-dependent | Points per cycle (5-100) |
 | `--max-cycles` | 10 | Maximum iterations (1-50) |
-| `--total-budget` | 500 | Cumulative points cap (5-5000) |
+| `--total-budget` | 500 | Cumulative point budget (5-5000). Advisory by default (tracked/surfaced, not a stop). |
+| `--enforce-budget` | off | Make `--total-budget` a hard stop again (also settable via config `budget.enforce: true`). |
 | `--continue` | — | Resume stopped/interrupted run |
 | `--stop` | — | Gracefully stop active run |
 | `--status` | — | Show current campaign progress |

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pan-wizard",
-  "version": "3.18.0",
+  "version": "3.19.0",
   "description": "Command a bot army for your codebase: an Opus Mission Control delegates whole-project goals to specialist squads and ships behind a human merge gate. Five AI CLIs, zero context rot.",
   "bin": {
     "pan-wizard": "bin/install.js"

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

@@ -76,6 +76,9 @@ function writeSchedule(cwd, opts, now) {
     source: opts.source ?? existing.source ?? 'backlog',
     cadence,
     daily_budget: opts.daily_budget != null ? Number(opts.daily_budget) : (existing.daily_budget ?? 300),
+    // Advisory by default: the daily budget is tracked + surfaced but does not block
+    // a due run unless explicitly enforced.
+    enforce_budget: opts.enforce_budget != null ? Boolean(opts.enforce_budget) : (existing.enforce_budget ?? false),
     enabled: opts.enabled != null ? Boolean(opts.enabled) : (existing.enabled ?? true),
     paused: opts.paused != null ? Boolean(opts.paused) : (existing.paused ?? false),
     next_due: existing.next_due ?? at.toISOString(),
@@ -108,7 +111,7 @@ function isRunDue(schedule, now) {
   const spent = spentToday(schedule, at);
   if (!schedule.enabled) return { due: false, reason: 'disabled', next_due: schedule.next_due, spent_today: spent };
   if (schedule.paused) return { due: false, reason: 'paused', next_due: schedule.next_due, spent_today: spent };
-  if (schedule.daily_budget != null && spent >= schedule.daily_budget) {
+  if (schedule.enforce_budget && schedule.daily_budget != null && spent >= schedule.daily_budget) {
     return { due: false, reason: 'budget_exhausted_today', next_due: schedule.next_due, spent_today: spent };
   }
   const due = new Date(schedule.next_due);

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

@@ -62,6 +62,10 @@ function buildConfigDefaults(hasBraveSearch, userDefaults) {
       default_points: 50,
       micro_threshold_tasks: 3,
       micro_threshold_files: 2,
+      // Budget is ADVISORY by default: point/spend caps are tracked and surfaced
+      // (HUD, telemetry) but never STOP a run. Set enforce:true (or pass
+      // --enforce-budget) to make the cap a hard stop again.
+      enforce: false,
     },
     commit: {
       safety_checks: true,

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

@@ -47,34 +47,44 @@ const COST_MULTIPLIERS = { reasoning: 15, mid: 3, fast: 1 };
 
 // ─── Model Profile Table ─────────────────────────────────────────────────────
 
+// COST RESET (2026-07): quality + balanced (the default) both resolve to the
+// `reasoning` tier for EVERY agent — i.e. the DEFAULT model you launched with
+// (`inherit`). PAN no longer silently demotes agents to cheaper models; context
+// isolation (each subagent runs in its own window), not a cheaper model, is what
+// keeps the main conversation clean. Cheapness is now OPT-IN: choose the `budget`
+// profile (the only column that still down-tiers) or pin a specific agent via
+// config `model_overrides`. The 3 security agents additionally pin `model: opus`
+// in their own frontmatter (a native, deliberate exception). resolve-model /
+// MODEL_PROFILES is advisory + cost-estimation; native Claude Code delegation
+// reads each agent file's static `model:` (unset → inherit).
 const MODEL_PROFILES = {
   // Original planning/execution agents (pre-v3.0)
   'pan-planner':              { quality: 'reasoning', balanced: 'reasoning', budget: 'mid' },
-  'pan-roadmapper':           { quality: 'reasoning', balanced: 'mid',      budget: 'mid' },
-  'pan-executor':             { quality: 'reasoning', balanced: 'mid',      budget: 'mid' },
-  'pan-phase-researcher':     { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-project-researcher':   { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-research-synthesizer': { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-debugger':             { quality: 'reasoning', balanced: 'mid',      budget: 'mid' },
-  'pan-document_code':        { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
-  'pan-verifier':             { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-plan-checker':         { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-integration-checker':  { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-reviewer':             { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
+  'pan-roadmapper':           { quality: 'reasoning', balanced: 'reasoning', budget: 'mid' },
+  'pan-executor':             { quality: 'reasoning', balanced: 'reasoning', budget: 'mid' },
+  'pan-phase-researcher':     { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-project-researcher':   { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-research-synthesizer': { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-debugger':             { quality: 'reasoning', balanced: 'reasoning', budget: 'mid' },
+  'pan-document_code':        { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-verifier':             { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-plan-checker':         { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-integration-checker':  { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-reviewer':             { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
   // Spec B v2 agents (v3.0–v3.4) — added v3.7.5 to close MODEL_PROFILES drift
   'pan-conductor':            { quality: 'reasoning', balanced: 'reasoning', budget: 'mid' },
-  'pan-counterfactual':       { quality: 'reasoning', balanced: 'mid',      budget: 'mid' },
-  'pan-hardener':             { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-meta-reviewer':        { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-knowledge':            { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-previewer':            { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
+  'pan-counterfactual':       { quality: 'reasoning', balanced: 'reasoning', budget: 'mid' },
+  'pan-hardener':             { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-meta-reviewer':        { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-knowledge':            { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-previewer':            { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
   // v3.5 agents
-  'pan-optimizer':            { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
-  'pan-distiller':            { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
+  'pan-optimizer':            { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
+  'pan-distiller':            { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
   // v3.7.0 self-improvement loop — observation-only watchdog
-  'pan-experiment-runner':    { quality: 'reasoning', balanced: 'fast',     budget: 'fast' },
+  'pan-experiment-runner':    { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
   // ADR-0033 bot-army — Release squad
-  'pan-release':              { quality: 'reasoning', balanced: 'mid',      budget: 'fast' },
+  'pan-release':              { quality: 'reasoning', balanced: 'reasoning', budget: 'fast' },
 };
 
 // ─── Effort Profiles (2026-06, adaptive-thinking era) ───────────────────────
@@ -279,7 +289,7 @@ function loadConfig(cwd) {
       verifier: get('verifier', { section: 'workflow', field: 'verifier' }) ?? defaults.verifier,
       parallelization,
       brave_search: get('brave_search') ?? defaults.brave_search,
-      budget: parsed.budget || { default_points: 50, micro_threshold_tasks: 3, micro_threshold_files: 2 },
+      budget: parsed.budget || { default_points: 50, micro_threshold_tasks: 3, micro_threshold_files: 2, enforce: false },
       commit: parsed.commit || { safety_checks: true, conventional_types: true, sensitive_patterns: ['\\.env$', '\\.pem$', '\\.key$', 'credentials', 'secret', 'password', 'token'] },
       execution: parsed.execution || { default_mode: 'wave_order', rollback_snapshots: true, error_pattern_learning: true },
       focus: parsed.focus || { auto_commit: true },

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

@@ -863,7 +863,9 @@ function focusAutoCheckpointCommit(cwd, cycle, run) {
 
 function determineStopReason(cycle, run) {
   if (cycle.tests_after < cycle.tests_before) return 'regression';
-  if (run.totals.points_used >= run.total_budget) return 'budget_cap';
+  // Budget is advisory by default — it only STOPS the run when explicitly enforced.
+  // Otherwise the overage is tracked/surfaced (indication) and the loop continues.
+  if (run.budget_enforce && run.totals.points_used >= run.total_budget) return 'budget_cap';
   if (run.totals.cycles_completed >= run.max_cycles) return 'max_cycles';
   if (cycle.items_completed === 0) {
     // Security category gets a descriptive stop reason rather than generic zero_completed
@@ -929,6 +931,10 @@ function focusAutoInit(cwd, raw, getVal, hasFlag) {
   const budget = Number(getVal('--budget', String(defaults.budget)));
   const maxCycles = Number(getVal('--max-cycles', String(DEFAULT_MAX_CYCLES)));
   const totalBudget = Number(getVal('--total-budget', String(DEFAULT_TOTAL_BUDGET)));
+  // Budget is advisory by default (tracked + surfaced, never a hard stop). Enforce
+  // only when the user opts in via config `budget.enforce` or `--enforce-budget`.
+  const budgetConfig = loadConfig(cwd).budget || {};
+  const budgetEnforce = hasFlag('--enforce-budget') || budgetConfig.enforce === true;
 
   if (!FOCUS_MODES.includes(mode)) return error(`Mode must be one of: ${FOCUS_MODES.join(', ')}`);
   if (budget < BUDGET_MIN || budget > BUDGET_MAX) return error(`Budget must be between ${BUDGET_MIN} and ${BUDGET_MAX}`);
@@ -947,6 +953,7 @@ function focusAutoInit(cwd, raw, getVal, hasFlag) {
     budget_per_cycle: budget,
     max_cycles: maxCycles,
     total_budget: totalBudget,
+    budget_enforce: budgetEnforce,
     priority_range: category ? CATEGORY_PRIORITY_RANGE[category] : { min: 0, max: 6 },
     deep_review_enabled: hasFlag('--deep-review'),
     tests_baseline: null,

package/pan-zcode/mcp/orchestrator.cjs

@@ -15,7 +15,9 @@
  * are reproducible.
  */
 
-const DEFAULT_CAPS = { maxCycles: 25, budget: Infinity };
+// Budget is advisory by default (enforceBudget:false) — it never stops the loop
+// unless the caller opts in. maxCycles remains a hard safety stop.
+const DEFAULT_CAPS = { maxCycles: 25, budget: Infinity, enforceBudget: false };
 const PHASE_NEXT = {
   none: 'plan',
   researched: 'plan',
@@ -47,7 +49,7 @@ function nextAction(state, caps) {
     return { action: 'stop', reason: 'regression', done: true };
   }
   if ((state.cycles || 0) >= c.maxCycles) return { action: 'stop', reason: 'max_cycles', done: true };
-  if ((state.points_used || 0) >= c.budget) return { action: 'stop', reason: 'budget_cap', done: true };
+  if (c.enforceBudget && (state.points_used || 0) >= c.budget) return { action: 'stop', reason: 'budget_cap', done: true };
 
   // The human merge gate is a barrier: while a merge awaits approval, do nothing else.
   if (state.awaiting_approval) return { action: 'await_approval', reason: 'human_gate', done: false };