@polderlabs/bizar 10.13.0 → 10.14.0

package/.claude/agents/debug-specialist.md

@@ -2,7 +2,7 @@
 name: carl
 description: Carl — VP Engineering. The ultimate fallback debugger. For the hardest problems when Karen stalls, when a bug has resisted 2+ attempts by lower-tier agents, or when novel root-cause insight is needed. Most expensive tier — use sparingly. Always plan → @linda audit → execute.
 tools: Read, Edit, Write, Bash, Glob, Grep, WebFetch, WebSearch, Agent, Skill
-model: cx/gpt-5.6-sol
+model: claude-qwen/qwen3.8-max
 isolation: worktree
 ---
 

package/.claude/agents/exec-assistant.md

@@ -2,7 +2,7 @@
 name: pam
 description: Pam — Executive Assistant. Fast single-shot agent for small edits, mechanical changes, one-shot questions. No delegation, no parallel streams, no Agent tool. Use for "rename this file", "fix this typo", quick lookups.
 tools: Read, Edit, Write, Bash, Glob, Grep, WebFetch, WebSearch, Skill
-model: bizar/MiniMax-M2.5
+model: claude-minimax/MiniMax-M2.5
 isolation: worktree
 ---
 

package/.claude/agents/help-desk.md

@@ -2,7 +2,7 @@
 name: susan
 description: Susan — Help Desk. Read-only codebase Q&A. Answers questions about the project with file:line references, never modifies anything. Use when the user asks "how does X work", "where is Y", "what's the architecture of Z".
 tools: Read, Glob, Grep, WebFetch, WebSearch, Skill
-model: bizar/MiniMax-M2.7
+model: claude-minimax/MiniMax-M2.7
 ---
 
 You are Susan, the front-desk Help Desk. You answer questions about the codebase. You never modify files. You never delegate. You explore and explain.

package/.claude/agents/it-lead.md

@@ -2,7 +2,7 @@
 name: steve
 description: Steve — IT Lead. Git and GitHub operations specialist. The only agent allowed to perform write-level git (commit, push, merge, rebase, branch) and `gh` CLI operations. Use for any commit, PR creation, GitHub issue, merge, or branch operation.
 tools: Read, Edit, Write, Bash, Glob, Grep, WebFetch, WebSearch, Agent, Skill
-model: bizar/MiniMax-M3
+model: claude-minimax/MiniMax-M3
 ---
 
 You are Steve, the IT Lead. You are the **only** agent allowed to perform write-level git operations. Every other agent forwards git work to you via Mike.

package/.claude/agents/knowledge-manager.md

@@ -2,7 +2,7 @@
 name: oscar
 description: Oscar — Knowledge Manager. Code search specialist using Semble. Find code by intent, locate implementations, understand how something works, discover related code. Prefer over Bash/Read/Grep for any semantic or exploratory question.
 tools: Read, Glob, Grep, WebFetch, WebSearch, Skill
-model: bizar/MiniMax-M2.7
+model: claude-minimax/MiniMax-M2.7
 ---
 
 You are Oscar, the Knowledge Manager. You are the code search specialist. You explore codebases semantically using Semble. You never modify anything. You return concise, file-referenced answers.

package/.claude/agents/office-coordinator.md

@@ -2,7 +2,7 @@
 name: brenda
 description: Brenda — Office Coordinator. Simple, routine, deterministic engineering tasks. Quick edits, mechanical work, file operations, and `.bizar/` maintenance after every implementation.
 tools: Read, Edit, Write, Bash, Glob, Grep, WebFetch, WebSearch, Skill
-model: bizar/MiniMax-M2.5
+model: claude-minimax/MiniMax-M2.5
 isolation: worktree
 ---
 

package/.claude/agents/office-greeter.md

@@ -2,7 +2,7 @@
 name: janet
 description: Janet — Office Greeter. Asks the one targeted, project-specific clarifying question that unblocks ambiguous or incomplete requests. Read-only, never implements. Use when Mike's incoming request is missing parameters, has multiple interpretations, or has contradictory constraints.
 tools: Read, Glob, Grep, WebFetch, WebSearch, AskUserQuestion, Skill
-model: bizar/MiniMax-M2.5
+model: claude-minimax/MiniMax-M2.5
 ---
 
 You are Janet, the Office Greeter. Mike calls on you when a request is ambiguous, incomplete, or has multiple reasonable interpretations. Your job: ask the one question that unblocks the work.

package/.claude/agents/office-manager.md

@@ -2,7 +2,7 @@
 name: mike
 description: Mike — Office Manager and the single main orchestrator. Routes every non-empty primary request, decomposes work, coordinates specialists, and synthesizes verified results without implementing.
 tools: Agent, Read, WebFetch, WebSearch
-model: cx/gpt-5.6-sol
+model: claude-qwen/qwen3.8-max
 ---
 
 You are Mike, the Office Manager. You NEVER execute work yourself. You analyze every request and delegate to subagents via the `Agent` tool (use `run_in_background: true` for async work). Your ONLY jobs: **decompose, route, synthesize**.
@@ -76,7 +76,7 @@ Run both in parallel via a single `Agent` message. Both are read-only; merge the
 
 Sequential — each step needs the previous output:
 
-1. **`@paul`** (premium, `cx/gpt-5.6-sol`) drafts the plan. Inputs: user's ask + Phase 1 findings. Output: 6-phase plan with file scopes.
+1. **`@paul`** (premium, `claude-qwen/qwen3.8-max`) drafts the plan. Inputs: user's ask + Phase 1 findings. Output: 6-phase plan with file scopes.
 2. **`@linda`** (high, `cx/gpt-5.6-terra`) audits adversarially:
    - `APPROVED` → proceed to Phase 3.
    - `CHANGES REQUIRED` → send corrections back to `@paul`, re-audit. Loop until clean.
@@ -204,7 +204,7 @@ When Todd and Karen both complete implementation work in parallel:
 
 ## Escalation — Route to @carl When Debug Stalls
 
-**Last-resort debug.** When a bug has resisted `@todd` and `@karen` for 2+ rounds, escalate to `@carl` (premium, `cx/gpt-5.6-sol`):
+**Last-resort debug.** When a bug has resisted `@todd` and `@karen` for 2+ rounds, escalate to `@carl` (premium, `claude-qwen/qwen3.8-max`):
 
 - Re-state the bug, the prior hypotheses tried, and what each ruled out.
 - Re-spawn `@greg` (parallel with `@oscar`) for a fresh targeted research pass if scope is wider than Carl can hold.

package/.claude/agents/planner.md

@@ -2,7 +2,7 @@
 name: paul
 description: Paul — Planning Specialist used by Mike after research. Produces phased, reversible plans with file scopes, Definition of Done, risks, and explicit stop conditions. Does not orchestrate, research initially, or implement.
 tools: Read, Glob, Grep, WebFetch, WebSearch, Skill, AskUserQuestion, Agent
-model: cx/gpt-5.6-sol
+model: claude-qwen/qwen3.8-max
 ---
 
 You are Paul, the Planning Specialist in Mike's planning phase. Mike is the

package/.claude/agents/research-analyst.md

@@ -2,7 +2,7 @@
 name: greg
 description: Repository and official-documentation researcher for Bizar plans and implementation work.
 tools: Read, Grep, Glob, WebFetch, WebSearch
-model: bizar/MiniMax-M3
+model: claude-minimax/MiniMax-M3
 ---
 
 # Greg — Research Analyst

package/.claude/agents/senior-engineer.md

@@ -2,7 +2,7 @@
 name: todd
 description: Todd — Senior Engineer. Mid-complexity implementation. New features, non-trivial debugging, refactoring, code review, and writing tests. Routes to the test gate after parallel implementation. Use for moderate-complexity implementation tasks.
 tools: Read, Edit, Write, Bash, Glob, Grep, WebFetch, WebSearch, Skill
-model: bizar/MiniMax-M2.7
+model: claude-minimax/MiniMax-M2.7
 isolation: worktree
 ---
 

package/.claude/agents/support-tech.md

@@ -2,7 +2,7 @@
 name: kevin
 description: Kevin — Support Tech. Read-only browser E2E verification with the official agent-browser CLI or MCP server.
 tools: Read, Bash, Glob, Grep, WebFetch, WebSearch, Skill
-model: bizar/MiniMax-M2.5
+model: claude-minimax/MiniMax-M2.5
 ---
 
 You are Kevin, the Support Tech. You verify user-facing web behavior with a

package/.claude/commands/use-default.md

@@ -1,14 +1,14 @@
 ---
-description: Revert Claude Code to the default tier (bizar/MiniMax-M3). Prints the exact shell command and env vars to unset.
+description: Revert Claude Code to the default tier (claude-minimax/MiniMax-M3). Prints the exact shell command and env vars to unset.
 allowed-tools: Bash
 ---
 
-# /use-default — Revert to bizar/MiniMax-M3 (default)
+# /use-default — Revert to claude-minimax/MiniMax-M3 (default)
 
 The local model router at `http://localhost:20128/v1` defaults to
-`bizar/MiniMax-M3` for everyday reasoning. Use this when you've
-been running on premium (`cx/gpt-5.6-sol`) and want to drop back to
-the cheaper default.
+`claude-minimax/MiniMax-M3` for everyday reasoning. Use this when you've
+been running on premium (`claude-qwen/qwen3.8-max`) and want to drop back
+to the cheaper default.
 
 ## Commands
 
@@ -30,7 +30,7 @@ claude
 ### Revert a project-level override
 
 If you added `ANTHROPIC_MODEL` to `.claude/settings.local.json` for
-this project, remove it (or set it back to `bizar/MiniMax-M3`).
+this project, remove it (or set it back to `claude-minimax/MiniMax-M3`).
 
 ```json
 {

package/.claude/commands/use-premium.md

@@ -1,17 +1,17 @@
 ---
-description: Launch Claude Code in this project on the premium tier (cx/gpt-5.6-sol). Prints the exact shell command and env vars to set.
+description: Launch Claude Code in this project on the premium tier (claude-qwen/qwen3.8-max). Prints the exact shell command and env vars to set.
 allowed-tools: Bash
 ---
 
-# /use-premium — Run this session on cx/gpt-5.6-sol (premium)
+# /use-premium — Run this session on claude-qwen/qwen3.8-max (premium)
 
 The local model router at `http://localhost:20128/v1` exposes
-`cx/gpt-5.6-sol` as the premium-tier reasoning model. Premium is
+`claude-qwen/qwen3.8-max` as the premium-tier reasoning model. Premium is
 expensive — use it intentionally, not by default.
 
 ## What it does
 
-Sets `ANTHROPIC_MODEL=cx/gpt-5.6-sol` and `ANTHROPIC_BASE_URL=http://localhost:20128/v1` so the entire session loop runs on the premium model. The model-router is the same one your subagents (`@paul`, `@ria`, `@brad`, `@linda`, `@carl`) already hit.
+Sets `ANTHROPIC_MODEL=claude-qwen/qwen3.8-max` and `ANTHROPIC_BASE_URL=http://localhost:20128/v1` so the entire session loop runs on the premium model. The model-router is the same one your subagents (`@mike`, `@paul`, `@carl`) already hit.
 
 ## Commands
 
@@ -19,7 +19,7 @@ Sets `ANTHROPIC_MODEL=cx/gpt-5.6-sol` and `ANTHROPIC_BASE_URL=http://localhost:2
 
 ```bash
 ANTHROPIC_BASE_URL=http://localhost:20128/v1 \
-ANTHROPIC_MODEL=cx/gpt-5.6-sol \
+ANTHROPIC_MODEL=claude-qwen/qwen3.8-max \
 claude
 ```
 
@@ -28,7 +28,7 @@ router as the model id):
 
 ```bash
 ANTHROPIC_BASE_URL=http://localhost:20128/v1 \
-claude --model cx/gpt-5.6-sol
+claude --model claude-qwen/qwen3.8-max
 ```
 
 ### Make it the default for this project
@@ -39,7 +39,7 @@ Add to your local override `.claude/settings.local.json`:
 {
   "env": {
     "ANTHROPIC_BASE_URL": "http://localhost:20128/v1",
-    "ANTHROPIC_MODEL": "cx/gpt-5.6-sol"
+    "ANTHROPIC_MODEL": "claude-qwen/qwen3.8-max"
   }
 }
 ```
@@ -49,9 +49,9 @@ Then every Claude Code session in this directory opens on premium.
 ### Available premium-tier models on the local router
 
 ```
-cx/gpt-5.6-sol    — strongest reasoning (planning, debug, design, audit)
-cx/gpt-5.6-terra  — high reasoning at lower cost (complex impl, routing)
-cx/gpt-5.6-luna   — visually-oriented mid-tier (design alt)
+claude-qwen/qwen3.8-max  — strongest reasoning (orchestration, planning, debug)
+cx/gpt-5.6-terra        — high reasoning at lower cost (complex impl, review)
+cx/gpt-5.6-luna         — visually-oriented mid-tier (design)
 ```
 
 See `.claude/model-router.json` for the full tier table.
@@ -65,10 +65,10 @@ See `.claude/model-router.json` for the full tier table.
 
 ## When NOT to use premium
 
-- Routine edits, mechanical work, single-file fixes (use default `bizar/MiniMax-M3`).
+- Routine edits, mechanical work, single-file fixes (use default `claude-minimax/MiniMax-M3`).
 - Research (always use `@greg` on default).
-- Implementation at moderate complexity (use `bizar/MiniMax-M2.7` via `@todd`).
+- Implementation at moderate complexity (use `claude-minimax/MiniMax-M2.7` via `@todd`).
 
-The harness's cost ceiling (`cost_ceiling_per_session_usd: 5.0` in
+The harness's cost ceiling (`costCeilingPerSessionUsd: 5.0` in
 `model-router.json`) will downgrade you automatically if the budget is
 exceeded mid-session.

package/.claude/model-router.json

@@ -1,8 +1,8 @@
 {
   "$schema": "https://bizar.dev/schema/model-router.v2.json",
-  "version": "11.0.0",
+  "version": "11.1.0",
   "endpoint": "http://localhost:20129/v1",
-  "comment": "Bizar assigns one exact gateway model to each shipped agent. Role selection and complexity/model selection are separate decisions; every run snapshots its assignments before dispatch.",
+  "comment": "Bizar assigns one exact gateway model to each shipped agent. Role selection and complexity/model selection are separate decisions; every run snapshots its assignments before dispatch. Premium orchestrator/debug/planning tier uses Qwen 3.8 Max via the 9router gateway; MiniMax tiers are routed as claude-minimax/* gateway IDs.",
   "gateway": {
     "required": true,
     "endpoint": "http://localhost:20129/v1",
@@ -13,8 +13,8 @@
   },
   "tiers": {
     "premium": {
-      "models": ["cx/gpt-5.6-sol"],
-      "purpose": "Primary orchestration, system planning, and last-resort debugging where reasoning quality dominates cost."
+      "models": ["claude-qwen/qwen3.8-max"],
+      "purpose": "Primary orchestration, system planning, and last-resort debugging where reasoning quality dominates cost. Qwen 3.8 Max is reserved for the user-facing orchestrator, planning, and Carl escalation."
     },
     "high": {
       "models": ["cx/gpt-5.6-terra"],
@@ -25,15 +25,15 @@
       "purpose": "Focused visual-system and interaction-design reasoning."
     },
     "default": {
-      "models": ["bizar/MiniMax-M3"],
+      "models": ["claude-minimax/MiniMax-M3"],
       "purpose": "Broad, high-throughput research and guarded delivery operations."
     },
     "mid": {
-      "models": ["bizar/MiniMax-M2.7"],
+      "models": ["claude-minimax/MiniMax-M2.7"],
       "purpose": "Bounded technical analysis, code search, and moderate implementation."
     },
     "budget": {
-      "models": ["bizar/MiniMax-M2.5"],
+      "models": ["claude-minimax/MiniMax-M2.5"],
       "purpose": "Deterministic edits, clarification, and mechanical verification."
     }
   },
@@ -85,19 +85,19 @@
   },
   "agents": {
     "mike": {
-      "model": "cx/gpt-5.6-sol",
+      "model": "claude-qwen/qwen3.8-max",
       "tier": "premium",
-      "rationale": "The sole main orchestrator must sustain cross-phase decomposition, coordination, and synthesis without losing project constraints."
+      "rationale": "The sole main orchestrator the user talks to; Qwen 3.8 Max sustains cross-phase decomposition, coordination, and synthesis without losing project constraints."
     },
     "paul": {
-      "model": "cx/gpt-5.6-sol",
+      "model": "claude-qwen/qwen3.8-max",
       "tier": "premium",
-      "rationale": "System-wide plans and reversible execution boundaries require the strongest long-horizon reasoning."
+      "rationale": "System-wide plans and reversible execution boundaries require the strongest long-horizon reasoning; shares the user-facing orchestrator tier."
     },
     "carl": {
-      "model": "cx/gpt-5.6-sol",
+      "model": "claude-qwen/qwen3.8-max",
       "tier": "premium",
-      "rationale": "Last-resort debugging is reserved for novel, cross-cutting failures after cheaper agents have stalled."
+      "rationale": "Last-resort debugging is reserved for novel cross-cutting failures; Qwen 3.8 Max gives Carl the same headroom as the user-facing orchestrator."
     },
     "karen": {
       "model": "cx/gpt-5.6-terra",
@@ -120,47 +120,47 @@
       "rationale": "UI critique, interaction detail, and accessibility review align with the focused design tier."
     },
     "greg": {
-      "model": "bizar/MiniMax-M3",
+      "model": "claude-minimax/MiniMax-M3",
       "tier": "default",
       "rationale": "Repository and official-documentation research needs broad context handling at high throughput."
     },
     "steve": {
-      "model": "bizar/MiniMax-M3",
+      "model": "claude-minimax/MiniMax-M3",
       "tier": "default",
       "rationale": "Guarded Git and delivery operations need dependable command reasoning more than speculative depth."
     },
     "todd": {
-      "model": "bizar/MiniMax-M2.7",
+      "model": "claude-minimax/MiniMax-M2.7",
       "tier": "mid",
       "rationale": "Moderate implementation is bounded enough for the mid tier while still requiring technical judgment."
     },
     "oscar": {
-      "model": "bizar/MiniMax-M2.7",
+      "model": "claude-minimax/MiniMax-M2.7",
       "tier": "mid",
       "rationale": "Intent-based code discovery is a bounded technical analysis task with concrete evidence checks."
     },
     "susan": {
-      "model": "bizar/MiniMax-M2.7",
+      "model": "claude-minimax/MiniMax-M2.7",
       "tier": "mid",
       "rationale": "Source-grounded project explanations need enough reasoning to connect code paths without implementation overhead."
     },
     "brenda": {
-      "model": "bizar/MiniMax-M2.5",
+      "model": "claude-minimax/MiniMax-M2.5",
       "tier": "budget",
       "rationale": "Routine deterministic edits and maintenance should use the lowest capable model."
     },
     "pam": {
-      "model": "bizar/MiniMax-M2.5",
+      "model": "claude-minimax/MiniMax-M2.5",
       "tier": "budget",
       "rationale": "Single-shot mechanical tasks are intentionally constrained to the budget model."
     },
     "janet": {
-      "model": "bizar/MiniMax-M2.5",
+      "model": "claude-minimax/MiniMax-M2.5",
       "tier": "budget",
       "rationale": "One targeted clarification is a narrow task that does not justify a higher reasoning tier."
     },
     "kevin": {
-      "model": "bizar/MiniMax-M2.5",
+      "model": "claude-minimax/MiniMax-M2.5",
       "tier": "budget",
       "rationale": "Scripted browser verification is mechanical and evidence-driven."
     }

package/.claude-plugin/plugin.json

@@ -2,7 +2,7 @@
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "bizar-harness",
   "displayName": "Bizar Harness",
-  "version": "10.13.0",
+  "version": "10.14.0",
   "description": "Guarded multi-agent workflows for Claude Code with a single orchestrator, durable workflow state, and explicit human approval boundaries.",
   "author": {
     "name": "Polderlabs"

package/cli/commands/model.mjs

@@ -16,7 +16,7 @@ const BIZAR_MODEL_ROUTER_URL = process.env.BIZAR_MODEL_ROUTER_URL
 const ANTHROPIC_AUTH_TOKEN = process.env.ANTHROPIC_AUTH_TOKEN || 'sk_9router';
 
 const PROVIDER_GROUPS = [
-  'cx/', 'bizar/', 'oc/', 'claude/', 'anthropic/',
+  'cx/', 'claude-minimax/', 'claude-qwen/', 'oc/', 'claude/', 'anthropic/',
 ];
 
 function showHelp() {
@@ -31,7 +31,7 @@ function showHelp() {
     Fetches GET ${BIZAR_MODEL_ROUTER_URL}/models?limit=1000
     and prints a table of provider | id | display_name.
     The /model picker inside Claude Code shows only "claude"/"anthropic" prefixed
-    IDs. This command exposes the full set (cx/, bizar/, oc/, etc.).
+    IDs. This command exposes the full set (cx/, claude-minimax/, claude-qwen/, oc/, etc.).
 
   Flags:
     --json   Emit { providers: { "cx/": [...], ... }, total: N }

package/cli/provision.mjs

@@ -932,7 +932,7 @@ export async function runProvision(opts = {}) {
   if (anyFail) console.log(chalk.yellow('  ⚠ Some steps had issues.'));
   else { console.log(chalk.bold.green('  ✓ Bizar is ready.')); console.log(chalk.dim('     Next: restart your Claude Code session.')); }
   console.log('');
-  console.log(chalk.dim('  Premium model: ANTHROPIC_MODEL=cx/gpt-5.6-sol claude'));
+  console.log(chalk.dim('  Premium model: ANTHROPIC_MODEL=claude-qwen/qwen3.8-max claude'));
   console.log(chalk.dim('  See /use-premium or .claude/commands/use-premium.md for the full launch snippet.'));
   console.log('');
   return { ok: !anyFail, mode: effectiveMode, state: detectState(), stepResults };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@polderlabs/bizar",
-  "version": "10.13.0",
+  "version": "10.14.0",
   "description": "Autonomous, human-in-the-loop multi-agent harness for Claude Code with guarded workflows, typed SDK primitives, and MCP tools.",
   "type": "module",
   "bin": {

package/packages/sdk/dist/version.d.ts

@@ -1,5 +1,5 @@
 /**
  * SDK version constant. Keep synchronized with the workspace package versions.
  */
-export declare const SDK_VERSION: "10.13.0";
+export declare const SDK_VERSION: "10.14.0";
 //# sourceMappingURL=version.d.ts.map

package/packages/sdk/dist/version.js

@@ -1,5 +1,5 @@
 /**
  * SDK version constant. Keep synchronized with the workspace package versions.
  */
-export const SDK_VERSION = "10.13.0";
+export const SDK_VERSION = "10.14.0";
 //# sourceMappingURL=version.js.map

package/packages/sdk/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@polderlabs/bizar-sdk",
-  "version": "10.13.0",
+  "version": "10.14.0",
   "type": "module",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",