muonroi-cli 1.8.3 → 1.8.4

package/LICENSE

@@ -1,21 +1,9 @@
-MIT License
-
-Copyright (c) 2026 muonroi
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+Copyright (c) 2026 muonroi. All rights reserved.
+
+Commercial / Proprietary License
+
+This software and associated documentation files (the "Software") are proprietary to muonroi.
+
+Unauthorized copying, modification, distribution, transmission, performance, display or other use of this Software, via any medium, is strictly prohibited.
+
+The Software is provided "as is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. In no event shall the authors or copyright holders be liable for any claim, damages or other liability, whether in an action of contract, tort or otherwise, arising from, out of or in connection with the software or the use or other dealings in the Software.

package/README.md

@@ -1,122 +1,133 @@
-<p align="center">
-  <h1 align="center">muonroi-cli</h1>
-  <p align="center">
-    <em>An AI coding agent where models argue with each other before answering.</em>
-  </p>
-  <p align="center">
-    <a href="https://github.com/muonroi/muonroi-cli/actions/workflows/ci-matrix.yml"><img alt="CI" src="https://github.com/muonroi/muonroi-cli/actions/workflows/ci-matrix.yml/badge.svg"></a>
-    <a href="https://www.npmjs.com/package/muonroi-cli"><img alt="npm" src="https://img.shields.io/npm/v/muonroi-cli.svg"></a>
-    <img alt="Providers" src="https://img.shields.io/badge/providers-7%20supported-blue">
-    <img alt="License" src="https://img.shields.io/badge/license-MIT-yellow">
-    <img alt="Runtime" src="https://img.shields.io/badge/runtime-Bun%201.3%2B-orange">
-  </p>
-</p>
-
----
-
-> Routes each task to the optimal model, runs adversarial multi-model debates for high-stakes decisions, and persists behavioral memory across sessions. Bring your own API keys. Total cost: ~$5/month.
-
-<p align="center">
-  <img src="https://raw.githubusercontent.com/muonroi/muonroi-cli/master/docs/demo.gif" alt="Council debate — REST vs gRPC decision" width="840" />
-</p>
-
-## Quick Start
-
-### Install
-
-**Recommended — prebuilt standalone binary (zero runtime deps, no prereqs):**
-
-Linux / macOS:
-```bash
-curl -fsSL https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.sh | bash
-```
-
-Windows PowerShell:
-```powershell
-irm https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.ps1 | iex
-```
-
-The installers download a `bun --compile` binary from GitHub Releases — single executable, all native deps bundled, no Node/Bun/build tools required.
-
-**Bun runtime (requires Bun ≥ 1.x):**
-```bash
-bun add -g muonroi-cli
-muonroi-cli
-```
-
-> **⚠ Why no `npm install -g`?** The TUI engine (`@opentui/core`) uses Bun-only `import ... with { type: "file" }` for Tree-sitter syntax highlight queries (`.scm` files). Node ESM cannot load these, so running under bare Node throws `ERR_UNKNOWN_FILE_EXTENSION`. Use the standalone installer (above) or `bun add -g`.
-
-### First run
-
-On first launch the CLI shows a wizard that:
-1. Lists supported providers (currently **DeepSeek** and **SiliconFlow**).
-2. Offers four ways to add credentials: paste an API key, import an encrypted bundle (`keys export`/`keys import`), sync from Bitwarden, or skip and add later via `/providers` inside the TUI.
-
-After setup, role routing auto-balances across enabled providers:
-
-```json
-// ~/.muonroi-cli/user-settings.json (generated by the wizard — edit if needed)
-{
-  "defaultProvider": "deepseek",
-  "providers": {
-    "deepseek":    { "enabled": true },
-    "siliconflow": { "enabled": true }
-  },
-  "roleModels": {
-    "leader":    "deepseek-v4-pro",
-    "implement": "deepseek-v4-flash",
-    "verify":    "deepseek-v4-pro",
-    "research":  "deepseek-v4-flash"
-  }
-}
-```
-
-### Moving keys between devices
-
-```bash
-# Source device — encrypts every stored key into one passphrase-protected file
-muonroi-cli keys export ~/muonroi-keys.json
-
-# Move the file via any channel (USB, Drive, AirDrop, email...)
-
-# Target device — same passphrase rehydrates the OS keychain
-muonroi-cli keys import ~/muonroi-keys.json
-```
-
-Inside the TUI press `/providers` then `B` to sync directly from a Bitwarden vault instead.
-
-### Updates
-
-The CLI checks npm once per day on startup and prompts when a newer version is available. Set `autoUpdate: true` in `user-settings.json` to skip the prompt and update silently. Manual: `muonroi-cli update`.
-
-## Documentation
-
-Full documentation at **[docs.muonroi.com/docs/cli](https://docs.muonroi.com/docs/cli/overview)**
-
-| Topic | Link |
-|---|---|
-| Overview & architecture | [CLI Overview](https://docs.muonroi.com/docs/cli/overview) |
-| Multi-Model Council | [Council Debate Guide](https://docs.muonroi.com/docs/cli/guides/council-debate) |
-| Prompt Intelligence Layer | [PIL Pipeline Guide](https://docs.muonroi.com/docs/cli/guides/pil-pipeline) |
-| Experience Engine | [Experience Engine Guide](https://docs.muonroi.com/docs/cli/guides/experience-engine) |
-| Agent Harness | [Agent Harness Guide](https://docs.muonroi.com/docs/cli/guides/agent-harness) |
-| Settings reference | [CLI Settings Reference](https://docs.muonroi.com/docs/cli/reference/cli-settings-reference) |
-| Commands reference | [Commands Reference](https://docs.muonroi.com/docs/cli/reference/commands-reference) |
-| Providers reference | [Providers Reference](https://docs.muonroi.com/docs/cli/reference/providers-reference) |
-
-## Development
-
-```bash
-git clone https://github.com/muonroi/muonroi-cli.git
-cd muonroi-cli && bun install
-
-bun run dev           # run from source
-bun run typecheck     # type check
-bun run test          # vitest
-bun run lint          # biome check
-bun run build:binary  # standalone binary
-```
-
-## License
-
-MIT
+<p align="center">
+  <h1 align="center">muonroi-cli</h1>
+  <p align="center">
+    <em>An AI coding agent where models argue with each other before answering.</em>
+  </p>
+  <p align="center">
+    <a href="https://github.com/muonroi/muonroi-cli/actions/workflows/ci-matrix.yml"><img alt="CI" src="https://github.com/muonroi/muonroi-cli/actions/workflows/ci-matrix.yml/badge.svg"></a>
+    <a href="https://www.npmjs.com/package/muonroi-cli"><img alt="npm" src="https://img.shields.io/npm/v/muonroi-cli.svg"></a>
+    <img alt="Providers" src="https://img.shields.io/badge/providers-7%20supported-blue">
+    <img alt="License" src="https://img.shields.io/badge/license-MIT-yellow">
+    <img alt="Runtime" src="https://img.shields.io/badge/runtime-Bun%201.3%2B-orange">
+  </p>
+</p>
+
+---
+
+> Routes each task to the optimal model, runs adversarial multi-model debates for high-stakes decisions, and persists behavioral memory across sessions. Bring your own API keys. Total cost: ~$5/month.
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/muonroi/muonroi-cli/master/docs/demo.gif" alt="Council debate — REST vs gRPC decision" width="840" />
+</p>
+
+## Quick Start
+
+### Install
+
+**Recommended — install globally via NPM (requires Node.js ≥ 20) or Bun (requires Bun ≥ 1.x):**
+
+Using NPM:
+```bash
+npm install -g muonroi-cli
+```
+
+Using Bun:
+```bash
+bun add -g muonroi-cli
+```
+
+Once installed, run:
+```bash
+muonroi-cli
+```
+
+**Alternative — Prebuilt standalone binary (requires access to the GitHub repository):**
+
+Linux / macOS:
+```bash
+# Note: Requires github authorization if the repository is private
+curl -fsSL https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.sh | bash
+```
+
+Windows PowerShell:
+```powershell
+# Note: Requires github authorization if the repository is private
+irm https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.ps1 | iex
+```
+
+The installers download a `bun --compile` binary from GitHub Releases — single executable, all native deps bundled, no Node/Bun/build tools required.
+
+### First run
+
+On first launch the CLI shows a wizard that:
+1. Lists supported providers (currently **DeepSeek** and **SiliconFlow**).
+2. Offers four ways to add credentials: paste an API key, import an encrypted bundle (`keys export`/`keys import`), sync from Bitwarden, or skip and add later via `/providers` inside the TUI.
+
+After setup, role routing auto-balances across enabled providers:
+
+```json
+// ~/.muonroi-cli/user-settings.json (generated by the wizard — edit if needed)
+{
+  "defaultProvider": "deepseek",
+  "providers": {
+    "deepseek":    { "enabled": true },
+    "siliconflow": { "enabled": true }
+  },
+  "roleModels": {
+    "leader":    "deepseek-v4-pro",
+    "implement": "deepseek-v4-flash",
+    "verify":    "deepseek-v4-pro",
+    "research":  "deepseek-v4-flash"
+  }
+}
+```
+
+### Moving keys between devices
+
+```bash
+# Source device — encrypts every stored key into one passphrase-protected file
+muonroi-cli keys export ~/muonroi-keys.json
+
+# Move the file via any channel (USB, Drive, AirDrop, email...)
+
+# Target device — same passphrase rehydrates the OS keychain
+muonroi-cli keys import ~/muonroi-keys.json
+```
+
+Inside the TUI press `/providers` then `B` to sync directly from a Bitwarden vault instead.
+
+### Updates
+
+The CLI checks npm once per day on startup and prompts when a newer version is available. Set `autoUpdate: true` in `user-settings.json` to skip the prompt and update silently. Manual: `muonroi-cli update`.
+
+## Documentation
+
+Full documentation at **[docs.muonroi.com/docs/cli](https://docs.muonroi.com/docs/cli/overview)**
+
+| Topic | Link |
+|---|---|
+| Overview & architecture | [CLI Overview](https://docs.muonroi.com/docs/cli/overview) |
+| Multi-Model Council | [Council Debate Guide](https://docs.muonroi.com/docs/cli/guides/council-debate) |
+| Prompt Intelligence Layer | [PIL Pipeline Guide](https://docs.muonroi.com/docs/cli/guides/pil-pipeline) |
+| Experience Engine | [Experience Engine Guide](https://docs.muonroi.com/docs/cli/guides/experience-engine) |
+| Agent Harness | [Agent Harness Guide](https://docs.muonroi.com/docs/cli/guides/agent-harness) |
+| Settings reference | [CLI Settings Reference](https://docs.muonroi.com/docs/cli/reference/cli-settings-reference) |
+| Commands reference | [Commands Reference](https://docs.muonroi.com/docs/cli/reference/commands-reference) |
+| Providers reference | [Providers Reference](https://docs.muonroi.com/docs/cli/reference/providers-reference) |
+
+## Development
+
+```bash
+git clone https://github.com/muonroi/muonroi-cli.git
+cd muonroi-cli && bun install
+
+bun run dev           # run from source
+bun run typecheck     # type check
+bun run test          # vitest
+bun run lint          # biome check
+bun run build:binary  # standalone binary
+```
+
+## License
+
+MIT

package/dist/src/cli/cost-forensics.js

@@ -46,21 +46,21 @@ export function collectCostForensics(sessionId) {
     const chain = getSessionChain(sessionId);
     const placeholders = chain.map(() => "?").join(",");
     const rows = db
-        .prepare(`
-    SELECT id, source, model, message_seq, input_tokens, output_tokens,
-           cache_read_tokens, cache_creation_tokens, cost_micros, created_at,
-           provider_options_shape
-    FROM usage_events
-    WHERE session_id IN (${placeholders})
-    ORDER BY id ASC
+        .prepare(`
+    SELECT id, source, model, message_seq, input_tokens, output_tokens,
+           cache_read_tokens, cache_creation_tokens, cost_micros, created_at,
+           provider_options_shape
+    FROM usage_events
+    WHERE session_id IN (${placeholders})
+    ORDER BY id ASC
   `)
         .all(...chain);
     const counts = db
-        .prepare(`
-    SELECT event_type, COUNT(*) AS c
-    FROM interaction_logs
-    WHERE session_id IN (${placeholders})
-    GROUP BY event_type
+        .prepare(`
+    SELECT event_type, COUNT(*) AS c
+    FROM interaction_logs
+    WHERE session_id IN (${placeholders})
+    GROUP BY event_type
   `)
         .all(...chain);
     const countMap = new Map(counts.map((r) => [r.event_type, r.c]));

package/dist/src/council/prompts.js

@@ -548,33 +548,33 @@ export function buildSynthesisPrompt(ctx) {
     }
     let extraContext = "";
     if (ctx.refineContext) {
-        extraContext += `
-## User Refinements
-${ctx.refineContext}
+        extraContext += `
+## User Refinements
+${ctx.refineContext}
 `;
     }
     if (ctx.planEmphasis) {
-        extraContext += `
-## Additional Instruction
-The user has requested a concrete action plan with executable steps. Each action item MUST be an object with these fields:
-  {
-    "step": "<imperative action>",
-    "owner_lens": "<which stance owns this — frontend / backend / architecture / etc>",
-    "time_estimate": "<rough — e.g. '2h', '1d', '~30min'>",
-    "depends_on": ["<step keys this requires>"] or [],
-    "acceptance_criteria": "<how we know it's done>"
-  }
-Order action items by dependency: predecessors first, dependents after.
-Risks MUST be objects with: {"description", "severity": "High|Medium|Low", "mitigation"}.
-Do NOT emit loose strings for these fields — the user needs structured plan output.
+        extraContext += `
+## Additional Instruction
+The user has requested a concrete action plan with executable steps. Each action item MUST be an object with these fields:
+  {
+    "step": "<imperative action>",
+    "owner_lens": "<which stance owns this — frontend / backend / architecture / etc>",
+    "time_estimate": "<rough — e.g. '2h', '1d', '~30min'>",
+    "depends_on": ["<step keys this requires>"] or [],
+    "acceptance_criteria": "<how we know it's done>"
+  }
+Order action items by dependency: predecessors first, dependents after.
+Risks MUST be objects with: {"description", "severity": "High|Medium|Low", "mitigation"}.
+Do NOT emit loose strings for these fields — the user needs structured plan output.
 `;
     }
     return {
         system,
-        prompt: `Final positions:
-${ctx.finalPositions}
-
-Full discussion:
+        prompt: `Final positions:
+${ctx.finalPositions}
+
+Full discussion:
 ${ctx.allExchanges}${extraContext}`,
     };
 }

package/dist/src/generated/version.d.ts

@@ -1,2 +1,2 @@
-export declare const PACKAGE_VERSION = "1.8.3";
+export declare const PACKAGE_VERSION = "1.8.4";
 export declare const PACKAGE_DESCRIPTION = "BYOK AI coding agent with multi-model council debate, role-based routing, and auto-compact.";

package/dist/src/generated/version.js

@@ -1,5 +1,5 @@
 // AUTO-GENERATED by scripts/sync-version.cjs. DO NOT EDIT BY HAND.
 // Sourced from package.json at build time so it survives bun --compile bundling.
-export const PACKAGE_VERSION = "1.8.3";
+export const PACKAGE_VERSION = "1.8.4";
 export const PACKAGE_DESCRIPTION = "BYOK AI coding agent with multi-model council debate, role-based routing, and auto-compact.";
 //# sourceMappingURL=version.js.map

package/dist/src/mcp/oauth-callback.js

@@ -1,7 +1,7 @@
 import http from "node:http";
 import { URL } from "node:url";
-const SUCCESS_HTML = `<!DOCTYPE html><html><body style="font-family:system-ui;text-align:center;padding:60px">
-<h2>Authorization successful</h2><p>You can close this tab and return to the terminal.</p>
+const SUCCESS_HTML = `<!DOCTYPE html><html><body style="font-family:system-ui;text-align:center;padding:60px">
+<h2>Authorization successful</h2><p>You can close this tab and return to the terminal.</p>
 </body></html>`;
 export function startOAuthCallbackServer(opts) {
     const callbackPath = opts.path ?? "/callback";

package/dist/src/mcp/setup-guide-text.d.ts

@@ -6,4 +6,4 @@
  * and the muonroi-tools MCP server (src/mcp/tools-server.ts, for external agents).
  * Keeping it here avoids duplicating ~70 lines across the two.
  */
-export declare const SETUP_GUIDE_TEXT = "# muonroi-cli Setup Guide\n\n## Install (zero runtime deps \u2014 recommended)\nLinux / macOS:\n  curl -fsSL https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.sh | bash\n\nWindows PowerShell:\n  irm https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.ps1 | iex\n\nBun (requires Bun >= 1.3):\n  bun add -g muonroi-cli\n\nNPM (requires Node >= 20):\n  npm install -g muonroi-cli\n\nThe installers fetch a pre-compiled single binary from GitHub Releases.\n\n## First run\n- Wizard appears automatically.\n- Lists supported providers (DeepSeek + SiliconFlow ready; others via BYOK).\n- Four credential options: paste key, Bitwarden sync (B in /providers), keys export/import (encrypted bundle), or skip for later.\n- Keys land in OS keychain (keytar). Settings written to ~/.muonroi-cli/user-settings.json.\n- Role routing (leader/implement/verify/research) is configured for you.\n\nAfter setup: run `muonroi-cli doctor` to validate.\n\n## Essential commands\n- Interactive TUI: `muonroi-cli` (or `node dist/index.js` after build)\n- Headless one-shot: `muonroi-cli --prompt \"your task\" --max-tool-rounds 8`\n- Health + MCP nudge: `muonroi-cli doctor`\n- Update: `muonroi-cli update` (or set \"autoUpdate\": true in user-settings)\n- Keys move between machines: `muonroi-cli keys export file.json` then import on target\n- Native tools MCP (for external agents): `muonroi-cli tools-mcp` (stdio)\n- Harness driver MCP: `muonroi-cli mcp-driver`\n\n## MCP integration (for Claude Desktop, Cursor, other agents)\nAdd to your MCP client config:\n\n{\n  \"mcpServers\": {\n    \"muonroi-tools\": {\n      \"command\": \"bun\",\n      \"args\": [\"run\", \"/absolute/path/to/muonroi-cli/src/index.ts\", \"tools-mcp\"]\n    }\n  }\n}\n\n(Use absolute path. After `bun run build`: \"node\", \"dist/index.js\", \"tools-mcp\")\n\nThe CLI's OWN inner agent exposes these as NATIVE in-process tools (no MCP self-spawn):\n- setup_guide (this document)\n- ee_query / ee_health / ee_feedback \u2014 Experience Engine semantic recall + compaction checkpoints + feedback for learning\n- usage_forensics <id-prefix> \u2014 per-session cost/token forensics (peak input, cache hits, anomalies)\n- lsp_query \u2014 goToDefinition, findReferences, hover, symbols, call hierarchy etc.\n- selfverify_* \u2014 Tier-1 heuristic + Tier-2 agentic self-QA harness runs (start/poll/result/cancel/list)\n\nFor BB/.NET template recipes and package docs, also connect an external \"muonroi-docs\" MCP server if available (provides docs_search + setup_guide for the templates).\n\n## Development\ngit clone https://github.com/muonroi/muonroi-cli.git\ncd muonroi-cli && bun install\n\nbun run dev                 # run from source (TUI)\nbun run typecheck           # tsc --noEmit\nbun run test                # vitest (unit + headless)\nbunx vitest -c vitest.harness.config.ts run tests/harness/   # TUI E2E (named-pipes on Win, fd3/4 on POSIX)\nbun run build               # or build:binary for standalone exe\n\nSee AGENTS.md (quick ref + rules), CLAUDE.md (harness verification), README.md.\n\n## Verify\nmuonroi-cli doctor\n# Checks runtimes, catalog load, keychain, MCP servers enabled, council research MCP nudge, EE reachability, recent error rate.\n# Any \"warn\" entries tell you exactly what to enable (e.g. tavily for web research in council).\n\nFor BB-aware scaffolding (/ideal on a muonroi-building-block target): ensure dotnet SDK + the three Muonroi.*.Template packages are installed via NuGet; doctor surfaces missing feed/template cases.\n";
+export declare const SETUP_GUIDE_TEXT = "# muonroi-cli Setup Guide\n\n## Install (Recommended \u2014 NPM/Bun global install)\nNPM (requires Node >= 20):\n  npm install -g muonroi-cli\n\nBun (requires Bun >= 1.3):\n  bun add -g muonroi-cli\n\n## Install (Alternative \u2014 Prebuilt standalone binary, requires repo access)\nLinux / macOS:\n  curl -fsSL https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.sh | bash\n\nWindows PowerShell:\n  irm https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.ps1 | iex\n\nThe installers fetch a pre-compiled single binary from GitHub Releases.\n\n## First run\n- Wizard appears automatically.\n- Lists supported providers (DeepSeek + SiliconFlow ready; others via BYOK).\n- Four credential options: paste key, Bitwarden sync (B in /providers), keys export/import (encrypted bundle), or skip for later.\n- Keys land in OS keychain (keytar). Settings written to ~/.muonroi-cli/user-settings.json.\n- Role routing (leader/implement/verify/research) is configured for you.\n\nAfter setup: run `muonroi-cli doctor` to validate.\n\n## Essential commands\n- Interactive TUI: `muonroi-cli` (or `node dist/index.js` after build)\n- Headless one-shot: `muonroi-cli --prompt \"your task\" --max-tool-rounds 8`\n- Health + MCP nudge: `muonroi-cli doctor`\n- Update: `muonroi-cli update` (or set \"autoUpdate\": true in user-settings)\n- Keys move between machines: `muonroi-cli keys export file.json` then import on target\n- Native tools MCP (for external agents): `muonroi-cli tools-mcp` (stdio)\n- Harness driver MCP: `muonroi-cli mcp-driver`\n\n## MCP integration (for Claude Desktop, Cursor, other agents)\nAdd to your MCP client config:\n\n{\n  \"mcpServers\": {\n    \"muonroi-tools\": {\n      \"command\": \"bun\",\n      \"args\": [\"run\", \"/absolute/path/to/muonroi-cli/src/index.ts\", \"tools-mcp\"]\n    }\n  }\n}\n\n(Use absolute path. After `bun run build`: \"node\", \"dist/index.js\", \"tools-mcp\")\n\nThe CLI's OWN inner agent exposes these as NATIVE in-process tools (no MCP self-spawn):\n- setup_guide (this document)\n- ee_query / ee_health / ee_feedback \u2014 Experience Engine semantic recall + compaction checkpoints + feedback for learning\n- usage_forensics <id-prefix> \u2014 per-session cost/token forensics (peak input, cache hits, anomalies)\n- lsp_query \u2014 goToDefinition, findReferences, hover, symbols, call hierarchy etc.\n- selfverify_* \u2014 Tier-1 heuristic + Tier-2 agentic self-QA harness runs (start/poll/result/cancel/list)\n\nFor BB/.NET template recipes and package docs, also connect an external \"muonroi-docs\" MCP server if available (provides docs_search + setup_guide for the templates).\n\n## Development\ngit clone https://github.com/muonroi/muonroi-cli.git\ncd muonroi-cli && bun install\n\nbun run dev                 # run from source (TUI)\nbun run typecheck           # tsc --noEmit\nbun run test                # vitest (unit + headless)\nbunx vitest -c vitest.harness.config.ts run tests/harness/   # TUI E2E (named-pipes on Win, fd3/4 on POSIX)\nbun run build               # or build:binary for standalone exe\n\nSee AGENTS.md (quick ref + rules), CLAUDE.md (harness verification), README.md.\n\n## Verify\nmuonroi-cli doctor\n# Checks runtimes, catalog load, keychain, MCP servers enabled, council research MCP nudge, EE reachability, recent error rate.\n# Any \"warn\" entries tell you exactly what to enable (e.g. tavily for web research in council).\n\nFor BB-aware scaffolding (/ideal on a muonroi-building-block target): ensure dotnet SDK + the three Muonroi.*.Template packages are installed via NuGet; doctor surfaces missing feed/template cases.\n";

package/dist/src/mcp/setup-guide-text.js

@@ -6,81 +6,82 @@
  * and the muonroi-tools MCP server (src/mcp/tools-server.ts, for external agents).
  * Keeping it here avoids duplicating ~70 lines across the two.
  */
-export const SETUP_GUIDE_TEXT = `# muonroi-cli Setup Guide
-
-## Install (zero runtime deps — recommended)
-Linux / macOS:
-  curl -fsSL https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.sh | bash
-
-Windows PowerShell:
-  irm https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.ps1 | iex
-
-Bun (requires Bun >= 1.3):
-  bun add -g muonroi-cli
-
-NPM (requires Node >= 20):
-  npm install -g muonroi-cli
-
-The installers fetch a pre-compiled single binary from GitHub Releases.
-
-## First run
-- Wizard appears automatically.
-- Lists supported providers (DeepSeek + SiliconFlow ready; others via BYOK).
-- Four credential options: paste key, Bitwarden sync (B in /providers), keys export/import (encrypted bundle), or skip for later.
-- Keys land in OS keychain (keytar). Settings written to ~/.muonroi-cli/user-settings.json.
-- Role routing (leader/implement/verify/research) is configured for you.
-
-After setup: run \`muonroi-cli doctor\` to validate.
-
-## Essential commands
-- Interactive TUI: \`muonroi-cli\` (or \`node dist/index.js\` after build)
-- Headless one-shot: \`muonroi-cli --prompt "your task" --max-tool-rounds 8\`
-- Health + MCP nudge: \`muonroi-cli doctor\`
-- Update: \`muonroi-cli update\` (or set "autoUpdate": true in user-settings)
-- Keys move between machines: \`muonroi-cli keys export file.json\` then import on target
-- Native tools MCP (for external agents): \`muonroi-cli tools-mcp\` (stdio)
-- Harness driver MCP: \`muonroi-cli mcp-driver\`
-
-## MCP integration (for Claude Desktop, Cursor, other agents)
-Add to your MCP client config:
-
-{
-  "mcpServers": {
-    "muonroi-tools": {
-      "command": "bun",
-      "args": ["run", "/absolute/path/to/muonroi-cli/src/index.ts", "tools-mcp"]
-    }
-  }
-}
-
-(Use absolute path. After \`bun run build\`: "node", "dist/index.js", "tools-mcp")
-
-The CLI's OWN inner agent exposes these as NATIVE in-process tools (no MCP self-spawn):
-- setup_guide (this document)
-- ee_query / ee_health / ee_feedback — Experience Engine semantic recall + compaction checkpoints + feedback for learning
-- usage_forensics <id-prefix> — per-session cost/token forensics (peak input, cache hits, anomalies)
-- lsp_query — goToDefinition, findReferences, hover, symbols, call hierarchy etc.
-- selfverify_* — Tier-1 heuristic + Tier-2 agentic self-QA harness runs (start/poll/result/cancel/list)
-
-For BB/.NET template recipes and package docs, also connect an external "muonroi-docs" MCP server if available (provides docs_search + setup_guide for the templates).
-
-## Development
-git clone https://github.com/muonroi/muonroi-cli.git
-cd muonroi-cli && bun install
-
-bun run dev                 # run from source (TUI)
-bun run typecheck           # tsc --noEmit
-bun run test                # vitest (unit + headless)
-bunx vitest -c vitest.harness.config.ts run tests/harness/   # TUI E2E (named-pipes on Win, fd3/4 on POSIX)
-bun run build               # or build:binary for standalone exe
-
-See AGENTS.md (quick ref + rules), CLAUDE.md (harness verification), README.md.
-
-## Verify
-muonroi-cli doctor
-# Checks runtimes, catalog load, keychain, MCP servers enabled, council research MCP nudge, EE reachability, recent error rate.
-# Any "warn" entries tell you exactly what to enable (e.g. tavily for web research in council).
-
-For BB-aware scaffolding (/ideal on a muonroi-building-block target): ensure dotnet SDK + the three Muonroi.*.Template packages are installed via NuGet; doctor surfaces missing feed/template cases.
+export const SETUP_GUIDE_TEXT = `# muonroi-cli Setup Guide
+
+## Install (Recommended — NPM/Bun global install)
+NPM (requires Node >= 20):
+  npm install -g muonroi-cli
+
+Bun (requires Bun >= 1.3):
+  bun add -g muonroi-cli
+
+## Install (Alternative — Prebuilt standalone binary, requires repo access)
+Linux / macOS:
+  curl -fsSL https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.sh | bash
+
+Windows PowerShell:
+  irm https://raw.githubusercontent.com/muonroi/muonroi-cli/master/install.ps1 | iex
+
+The installers fetch a pre-compiled single binary from GitHub Releases.
+
+## First run
+- Wizard appears automatically.
+- Lists supported providers (DeepSeek + SiliconFlow ready; others via BYOK).
+- Four credential options: paste key, Bitwarden sync (B in /providers), keys export/import (encrypted bundle), or skip for later.
+- Keys land in OS keychain (keytar). Settings written to ~/.muonroi-cli/user-settings.json.
+- Role routing (leader/implement/verify/research) is configured for you.
+
+After setup: run \`muonroi-cli doctor\` to validate.
+
+## Essential commands
+- Interactive TUI: \`muonroi-cli\` (or \`node dist/index.js\` after build)
+- Headless one-shot: \`muonroi-cli --prompt "your task" --max-tool-rounds 8\`
+- Health + MCP nudge: \`muonroi-cli doctor\`
+- Update: \`muonroi-cli update\` (or set "autoUpdate": true in user-settings)
+- Keys move between machines: \`muonroi-cli keys export file.json\` then import on target
+- Native tools MCP (for external agents): \`muonroi-cli tools-mcp\` (stdio)
+- Harness driver MCP: \`muonroi-cli mcp-driver\`
+
+## MCP integration (for Claude Desktop, Cursor, other agents)
+Add to your MCP client config:
+
+{
+  "mcpServers": {
+    "muonroi-tools": {
+      "command": "bun",
+      "args": ["run", "/absolute/path/to/muonroi-cli/src/index.ts", "tools-mcp"]
+    }
+  }
+}
+
+(Use absolute path. After \`bun run build\`: "node", "dist/index.js", "tools-mcp")
+
+The CLI's OWN inner agent exposes these as NATIVE in-process tools (no MCP self-spawn):
+- setup_guide (this document)
+- ee_query / ee_health / ee_feedback — Experience Engine semantic recall + compaction checkpoints + feedback for learning
+- usage_forensics <id-prefix> — per-session cost/token forensics (peak input, cache hits, anomalies)
+- lsp_query — goToDefinition, findReferences, hover, symbols, call hierarchy etc.
+- selfverify_* — Tier-1 heuristic + Tier-2 agentic self-QA harness runs (start/poll/result/cancel/list)
+
+For BB/.NET template recipes and package docs, also connect an external "muonroi-docs" MCP server if available (provides docs_search + setup_guide for the templates).
+
+## Development
+git clone https://github.com/muonroi/muonroi-cli.git
+cd muonroi-cli && bun install
+
+bun run dev                 # run from source (TUI)
+bun run typecheck           # tsc --noEmit
+bun run test                # vitest (unit + headless)
+bunx vitest -c vitest.harness.config.ts run tests/harness/   # TUI E2E (named-pipes on Win, fd3/4 on POSIX)
+bun run build               # or build:binary for standalone exe
+
+See AGENTS.md (quick ref + rules), CLAUDE.md (harness verification), README.md.
+
+## Verify
+muonroi-cli doctor
+# Checks runtimes, catalog load, keychain, MCP servers enabled, council research MCP nudge, EE reachability, recent error rate.
+# Any "warn" entries tell you exactly what to enable (e.g. tavily for web research in council).
+
+For BB-aware scaffolding (/ideal on a muonroi-building-block target): ensure dotnet SDK + the three Muonroi.*.Template packages are installed via NuGet; doctor surfaces missing feed/template cases.
 `;
 //# sourceMappingURL=setup-guide-text.js.map

package/dist/src/ops/doctor.js

@@ -169,9 +169,9 @@ async function checkBrainEmptiness() {
         // Count ee_injection events with event_subtype='no_match' in last 30 days
         const cutoff = new Date(Date.now() - 30 * 86_400_000).toISOString();
         const row = db
-            .prepare(`SELECT COUNT(*) as cnt FROM interaction_logs
-         WHERE event_type = 'ee_injection'
-           AND event_subtype = 'no_match'
+            .prepare(`SELECT COUNT(*) as cnt FROM interaction_logs
+         WHERE event_type = 'ee_injection'
+           AND event_subtype = 'no_match'
            AND created_at >= ?`)
             .get(cutoff);
         const noMatchCount = row?.cnt ?? 0;
@@ -293,10 +293,10 @@ async function checkCouncilMcpNudge() {
         // 2. Query DB for [Council Memory] records with URL or research topics
         const db = getDatabase();
         const rows = db
-            .prepare(`SELECT message_json FROM messages
-         WHERE role = 'system'
-           AND message_json LIKE '%[Council Memory]%'
-         ORDER BY created_at DESC
+            .prepare(`SELECT message_json FROM messages
+         WHERE role = 'system'
+           AND message_json LIKE '%[Council Memory]%'
+         ORDER BY created_at DESC
          LIMIT 50`)
             .all();
         let qualifyingCount = 0;