@polderlabs/bizar 3.9.0 → 3.9.1

package/config/AGENTS.md

@@ -102,6 +102,7 @@ BizarHarness ships always-on coding rules organized by language and concern. All
 | `rules/python.md` | Python conventions |
 | `rules/git.md` | Git and commit conventions |
 | `rules/testing.md` | Test methodology and coverage |
+| `rules/thinking.md` | All agents — concise thinking behavior |
 
 ### How to Use
 
@@ -110,6 +111,10 @@ BizarHarness ships always-on coding rules organized by language and concern. All
 3. All implementation agents (Heimdall, Thor, Tyr, Vidarr) MUST follow these rules
 4. Agents MAY propose additions to the rule files when patterns are discovered
 
+### Thinking Rule
+
+For agents with `reasoning: true` + `variant: "high"`, follow `rules/thinking.md` strictly. Cap reasoning at 2–4 sentences. No informal self-talk, no "what if" loops, no mid-thought self-correction. Think once, decide, act.
+
 ---
 
 ## Model Routing & Agents

package/config/agents/baldr.md

@@ -184,6 +184,9 @@ You may be dispatched alongside sibling agents working on the same repository at
 ### Reporting
 End your final summary with: `Siblings: <list>. Conflicts: <list or "none">. Git ops performed: <list or "none">.`
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ---
 
 ## Always-On Behavior Baseline

package/config/agents/forseti.md

@@ -124,6 +124,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Parallel Execution
 
 You may be invoked alongside other audit agents (parallel reviews of different files) or alongside implementation agents. The shared `AGENTS.md` baseline rules apply.

package/config/agents/frigg.md

@@ -111,6 +111,10 @@ The injected message you will see is exactly one of:
 - `[loop guard: 8 identical calls to <tool>]. Consider using the task tool to report back to your parent with what you've learned and what you need.`
 - An error containing: `Loop protection: 12 identical calls to <tool>. Use task to escalate.`
 
+
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ---
 
 ## Always-On Behavior Baseline

package/config/agents/heimdall.md

@@ -169,6 +169,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Parallel Execution
 
 You may be dispatched alongside sibling agents working on the same repository at the same time. The shared `AGENTS.md` baseline contains the universal rules — read those first. This section adds role-specific guidance.

package/config/agents/hermod.md

@@ -144,6 +144,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## PR Review Mode
 
 When dispatched for a `/pr-review`:

package/config/agents/mimir.md

@@ -127,6 +127,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Parallel Execution
 
 You may be dispatched alongside sibling agents working on the same repository at the same time. The shared `AGENTS.md` baseline contains the universal rules — read those first. This section adds role-specific guidance.

package/config/agents/quick.md

@@ -90,6 +90,10 @@ The injected message you will see is exactly one of:
 - `[loop guard: 8 identical calls to <tool>]. Consider using the task tool to report back to your parent with what you've learned and what you need.`
 - An error containing: `Loop protection: 12 identical calls to <tool>. Use task to escalate.`
 
+
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ---
 
 ## Always-On Behavior Baseline

package/config/agents/semble-search.md

@@ -50,6 +50,9 @@ If `semble` is not on `$PATH`, use `uvx --from "semble[mcp]" semble` in its plac
 
 ---
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Always-On Behavior Baseline
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.

package/config/agents/thor.md

@@ -109,6 +109,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Parallel Execution
 
 You may be dispatched alongside sibling agents working on the same repository at the same time. The shared `AGENTS.md` baseline contains the universal rules — read those first. This section adds role-specific guidance.

package/config/agents/tyr.md

@@ -108,6 +108,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Parallel Execution
 
 You may be dispatched alongside sibling agents working on the same repository at the same time. The shared `AGENTS.md` baseline contains the universal rules — read those first. This section adds role-specific guidance.

package/config/agents/vidarr.md

@@ -112,6 +112,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ## Parallel Execution
 
 You may be dispatched alongside sibling agents working on the same repository at the same time. The shared `AGENTS.md` baseline contains the universal rules — read those first. This section adds role-specific guidance.

package/config/agents/vor.md

@@ -150,6 +150,9 @@ Be professional and concise. Do not write long essays for every action.
 - One sentence of context beats three paragraphs of preamble.
 - Match the user's register: if they write briefly, reply briefly. If they want depth, they will ask.
 
+## Thinking style
+Follow `config/rules/thinking.md` strictly. Be precise, concise, and decisive in reasoning. No informal self-talk, no "what if" loops, no mid-thought self-correction.
+
 ---
 
 ## Always-On Behavior Baseline

package/config/opencode.json

@@ -0,0 +1,368 @@
+{
+  "$schema": "https://opencode.ai/config.json",
+  "model": "minimax/MiniMax-M3",
+  "small_model": "minimax/MiniMax-M2.7",
+  "default_agent": "odin",
+  "permission": "allow",
+  "snapshot": false,
+  "mcp": {
+    "supabase": {
+      "type": "remote",
+      "url": "https://mcp.supabase.com/mcp",
+      "enabled": false
+    },
+    "hindsight": {
+      "type": "remote",
+      "url": "https://memory-api.polderlabs.io/mcp",
+      "enabled": false,
+      "oauth": false,
+      "headers": {
+        "Authorization": "Bearer __HINDSIGHT_BEARER_TOKEN__",
+        "Content-Type": "application/json"
+      }
+    },
+    "semble": {
+      "type": "local",
+      "command": [
+        "uvx",
+        "--from",
+        "semble[mcp]",
+        "semble"
+      ],
+      "enabled": true
+    }
+  },
+  "plugin": [
+    ["./plugins/bizar/index.ts", {
+      "loopThresholdWarn": 5,
+      "loopThresholdEscalate": 8,
+      "loopThresholdBlock": 12,
+      "loopWindowSize": 10
+    }]
+  ],
+  "tools": {
+    "bizar_plan_action": true,
+    "bizar_get_plan_comments": true,
+    "bizar_wait_for_feedback": true,
+    "bizar_spawn_background": true,
+    "bizar_status": true,
+    "bizar_collect": true,
+    "bizar_kill": true
+  },
+  "agent": {
+    "odin": {
+      "description": "Odin — Pure router that delegates all work to subagents. Routes across Frigg (DeepSeek/Q&A), Vör (DeepSeek/clarify), Mimir (DeepSeek/research), Heimdall (DeepSeek/simple), Hermod (M2.7/git), Thor (M2.7/mid), Baldr (M2.7/design), Tyr (M3/top), Vidarr (GPT-5.5/ultra), Forseti (verifier/M3).",
+      "mode": "primary",
+      "model": "minimax/MiniMax-M3",
+      "color": "#6366f1",
+      "permission": {
+        "task": "allow",
+        "read": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow",
+        "hindsight_list_banks": "allow",
+        "hindsight_create_bank": "allow"
+      }
+    },
+    "vor": {
+      "description": "Vör — The Questioning One. When a task is ambiguous, incomplete, or unclear, Vör asks the right clarifying questions before any work begins.",
+      "mode": "subagent",
+      "model": "opencode/deepseek-v4-flash-free",
+      "color": "#8b5cf6",
+      "permission": {
+        "read": "allow",
+        "list": "allow",
+        "question": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "frigg": {
+      "description": "Frigg — All-knowing Q&A agent. Read-only codebase questions and answers. Never edits, never writes, only answers.",
+      "mode": "primary",
+      "model": "opencode/deepseek-v4-flash-free",
+      "color": "#06b6d4",
+      "permission": {
+        "read": "allow",
+        "list": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "bash": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow",
+        "question": "allow"
+      }
+    },
+    "quick": {
+      "description": "Quick — Fast single-shot tasks. No delegation, no parallel streams. Use for small edits, mechanical changes, one-shot questions. Routes to no one.",
+      "mode": "primary",
+      "model": "minimax/MiniMax-M2.7",
+      "color": "#22d3ee",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "write": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "task": "deny",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "mimir": {
+      "description": "Mimir — Dedicated research and codebase exploration agent. Uses Semble as primary search tool. Deep codebase analysis, pattern discovery, and documentation research.",
+      "mode": "subagent",
+      "model": "opencode/deepseek-v4-flash-free",
+      "color": "#0ea5e9",
+      "permission": {
+        "read": "allow",
+        "write": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "todowrite": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "heimdall": {
+      "description": "Heimdall — Simple, routine, and deterministic tasks using DeepSeek. Quick edits, mechanical work, file operations. The ever-watchful guardian.",
+      "mode": "subagent",
+      "model": "opencode/deepseek-v4-flash-free",
+      "color": "#10b981",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "hermod": {
+      "description": "Hermod — Git and GitHub operations specialist using MiniMax M2.7. Branching, commits, PRs, merge/rebase, conflict resolution, CI/CD, releases, gh CLI.",
+      "mode": "subagent",
+      "model": "minimax/MiniMax-M2.7",
+      "color": "#06b6d4",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "thor": {
+      "description": "Thor — Handles medium-complexity tasks using MiniMax M2.7 from minimax.io. Strong and reliable, cheaper than Tyr but more capable than Heimdall.",
+      "mode": "subagent",
+      "model": "minimax/MiniMax-M2.7",
+      "color": "#a855f7",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "baldr": {
+      "description": "Baldr — UI/UX design system specialist. Creates DESIGN.md files using Google's design.md standard. Focuses on visual consistency, usability, accessibility, and design tokens.",
+      "mode": "subagent",
+      "model": "minimax/MiniMax-M2.7",
+      "color": "#ec4899",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "tyr": {
+      "description": "Tyr — Handles the most complex implementation, debugging, and architectural work using MiniMax M3 via minimax.io. Unmatched wisdom for the hardest problems.",
+      "mode": "subagent",
+      "model": "minimax/MiniMax-M3",
+      "color": "#f59e0b",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "vidarr": {
+      "description": "Vidarr — The ultimate fallback using GPT-5.5 via OpenAI ChatGPT subscription. For the hardest problems when debugging stalls or nothing else works. Use sparingly — highest cost.",
+      "mode": "subagent",
+      "model": "openai/gpt-5.5",
+      "color": "#dc2626",
+      "permission": {
+        "read": "allow",
+        "edit": "allow",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "forseti": {
+      "description": "Forseti — Audits, criticizes, and corrects implementation plans before execution using MiniMax M3. No write permissions — review only.",
+      "mode": "subagent",
+      "model": "minimax/MiniMax-M3",
+      "color": "#ef4444",
+      "permission": {
+        "read": "allow",
+        "edit": "deny",
+        "bash": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow",
+        "todowrite": "allow",
+        "question": "allow",
+        "webfetch": "allow",
+        "websearch": "allow",
+        "hindsight_recall": "allow",
+        "hindsight_retain": "allow"
+      }
+    },
+    "semble-search": {
+      "description": "Code search agent for exploring any codebase. Use for finding code by intent, locating implementations, understanding how something works.",
+      "mode": "subagent",
+      "model": "opencode/deepseek-v4-flash-free",
+      "color": "#0ea5e9",
+      "permission": {
+        "bash": "allow",
+        "read": "allow",
+        "glob": "allow",
+        "grep": "allow",
+        "list": "allow"
+      }
+    }
+  },
+  "command": {
+    "audit": {
+      "description": "Run bizar audit to scan agent configuration for security issues.",
+      "agent": "forseti",
+      "template": "commands/audit.md"
+    },
+    "explain": {
+      "description": "Route to @frigg for read-only codebase Q&A. She will explore the code and answer without making any changes.",
+      "agent": "frigg",
+      "template": "commands/explain.md"
+    },
+    "init": {
+      "description": "Run bizar init to detect project stack, install relevant skills, and create .bizar/PROJECT.md.",
+      "agent": "heimdall",
+      "template": "commands/init.md"
+    },
+    "learn": {
+      "description": "Route to @heimdall. Extract patterns from the current session and append to .bizar/AGENTS_SELF_IMPROVEMENT.md.",
+      "agent": "heimdall",
+      "template": "commands/learn.md"
+    },
+    "plan": {
+      "description": "Open the Bizar visual plan canvas or manage existing plans.",
+      "agent": "odin",
+      "template": "commands/plan.md"
+    },
+    "pr-review": {
+      "description": "Route to @hermod for PR review mode. Launches @mimir (research) and @forseti (audit) in parallel, then posts the combined review as a PR comment.",
+      "agent": "hermod",
+      "template": "commands/pr-review.md"
+    },
+    "tailscale-serve": {
+      "description": "Authenticate Tailscale Serve and expose a local port on your tailnet.",
+      "agent": "heimdall",
+      "template": "commands/tailscale-serve.md"
+    },
+    "visual-plan": {
+      "description": "Toggle the Bizar visual plan canvas on or off, or check its current status.",
+      "agent": "odin",
+      "template": "commands/visual-plan.md"
+    },
+    "bizar": {
+      "description": "Bizar Plugin Menu — route to the right Bizar action based on the user's request.",
+      "agent": "odin",
+      "template": "commands/bizar.md",
+      "arguments": "$ARGUMENTS"
+    }
+  },
+  "provider": {
+    "minimax": {
+      "models": {
+        "MiniMax-M3": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        },
+        "MiniMax-M2.7": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        }
+      }
+    },
+    "openrouter": {
+      "models": {
+        "minimax/minimax-m3": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        },
+        "minimax/minimax-m2.7": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        },
+        "minimax/owl-alpha": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        }
+      }
+    }
+  }
+}

package/config/opencode.json.template

@@ -319,5 +319,35 @@
       "agent": "odin",
       "template": "commands-bizar/bizar.md\n\n$ARGUMENTS"
     }
+  },
+  "provider": {
+    "minimax": {
+      "models": {
+        "MiniMax-M3": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        },
+        "MiniMax-M2.7": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        }
+      }
+    },
+    "openrouter": {
+      "models": {
+        "minimax/minimax-m3": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        },
+        "minimax/minimax-m2.7": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        },
+        "minimax/owl-alpha": {
+          "interleaved": { "field": "reasoning_details" },
+          "reasoning": true
+        }
+      }
+    }
   }
 }

package/config/rules/thinking.md

@@ -0,0 +1,56 @@
+# Concise Thinking Rule
+
+## The Problem
+
+Models with `reasoning: true` + `variant: "high"` produce unbounded self-talk: "oh but what if this... actually let me reconsider... wait..." This is verbose, indecisive, and costly.
+
+## The Rule
+
+**Think in 2–4 sentences. One-shot. No loops.**
+
+### Structure
+
+Every reasoning output must follow this shape:
+
+```
+[Premise — what I know] → [Decision — what I'm doing] → [Next step]
+```
+
+No hedging. No "but what if". No mid-thought self-correction.
+
+### Hard Bans
+
+Never use these phrases in thinking output:
+- "oh but what if..."
+- "actually..."
+- "let me think..."
+- "wait..."
+- "hmm..."
+- "I wonder..."
+- "on second thought..."
+- "alternatively..."
+
+### One-Shot Pattern
+
+Think once. Decide. Act. Move on.
+
+**Forbidden patterns:**
+- "Option A... no wait Option B... actually let's try C..."
+- "I considered X, but then I realized Y, so maybe Z..."
+- "Let me reconsider..." followed by another reconsideration
+
+### What Thinking Should Look Like
+
+**BAD:**
+> "The file is missing a semicolon. Oh but what if it's actually a parsing issue? Let me think... Actually, looking more carefully, the real problem is the import statement is wrong. Wait, I should also check if the export is correct. Actually let me try fixing the import first..."
+
+**GOOD:**
+> "The import path is wrong — `../utils` should be `./utils`. Fixing the path resolves the missing export error. Next: re-run the build."
+
+### Length Cap
+
+Maximum **80 words** of thinking output before the first tool call or response. If you cannot decide in 80 words, state the decision you reached and proceed.
+
+### Enforcement
+
+This rule is injected into every model that supports `reasoning: true`. Non-compliance produces verbose, unhelpful output and is a correctness failure.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@polderlabs/bizar",
-  "version": "3.9.0",
+  "version": "3.9.1",
   "description": "Norse-pantheon multi-agent system for opencode — 13 agents across 4 cost tiers with cost-aware routing, plans, and a configurable agent harness.",
   "type": "module",
   "bin": {
@@ -42,5 +42,13 @@
   },
   "devDependencies": {
     "typescript": "^5.0.0"
-  }
+  },
+  "repository": {
+    "type": "git",
+    "url": "git@github.com:DrB0rk/BizarHarness.git"
+  },
+  "bugs": {
+    "url": "https://github.com/DrB0rk/BizarHarness/issues"
+  },
+  "homepage": "https://github.com/DrB0rk/BizarHarness#readme"
 }