MCPInstall

codexmonitor-orchestration

by reason-machinesCode Quality & Workflow

Expert in CodexMonitor, a Tauri app for orchestrating multiple Codex agents across local workspaces with threads, git integration, and remote daemon support.

Installs
265
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/reason-machines/codex-skills --skill codexmonitor-orchestration
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads codexmonitor-orchestration automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

CodexMonitor Orchestration Skill

Skill by ara.so — Codex Skills collection.

CodexMonitor is a Tauri-based desktop and mobile app for orchestrating multiple Codex agents across local workspaces. It provides workspace management, thread persistence, git/GitHub integration, file browsing, prompt libraries, and a remote daemon mode for connecting iOS clients or headless setups.

What CodexMonitor Does

  • Multi-workspace orchestration: Spawn one codex app-server per workspace, resume threads, track unread/running state
  • Thread management: Pin, rename, archive, copy threads; per-thread drafts; stop/interrupt in-flight turns
  • Worktree agents: Clone agents for isolated work under app data directory (legacy .codex-worktrees supported)
  • Git & GitHub: Diff stats, staged/unstaged files, commit log, branch management, GitHub Issues/PRs via gh
  • Composer: Image attachments, autocomplete for skills ($), prompts (/prompts:), reviews (/review), file paths (@)
  • Remote daemon: Run Codex on another machine, connect iOS client via TCP (Tailscale support)
  • Prompt library: Global/workspace prompts with create/edit/delete/move and run in threads
  • File tree: Search, file-type icons, reveal in Finder/Explorer
  • Terminal dock: Multiple tabs for background commands (experimental)

Installation

Requirements

  • Node.js + npm
  • Rust toolchain (stable)
  • CMake (for native dependencies, dictation/Whisper)
  • LLVM/Clang (Windows only, for bindgen)
  • Codex CLI installed and in PATH
  • Git CLI (for worktree operations)
  • GitHub CLI gh (optional, for GitHub integrations)

Install Dependencies

npm install

Check Environment

npm run doctor

Run in Development

npm run tauri:dev

Build Production Bundle

# macOS/Linux
npm run tauri:build

# Windows (opt-in, uses separate config)
npm run tauri:build:win

Artifacts: src-tauri/target/release/bundle/ (platform-specific subfolders)

Workspace Management

Adding a Workspace

Workspaces persist to workspaces.json in app data directory.

Via UI: Sidebar → Add workspace → Select directory

Data structure (src-tauri/src/workspaces/mod.rs):

#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct Workspace {
    pub id: String,
    pub name: String,
    pub path: String,
    pub codex_home: Option<String>, // Overrides global Codex home
    pub is_remote: bool,
    pub remote_host: Option<String>,
    pub remote_token: Option<String>,
}

Programmatic Workspace Access

Frontend service (src/services/tauri.ts):

import { invoke } from '@tauri-apps/api/tauri';

// Load all workspaces
const workspaces = await invoke<Workspace[]>('workspace_list');

// Add workspace
const newWorkspace = await invoke<Workspace>('workspace_add', {
  path: '/path/to/project',
  name: 'My Project',
});

// Update workspace
await invoke('workspace_update', {
  id: 'workspace-id',
  updates: { codex_home: '/custom/codex/home' },
});

// Remove workspace
await invoke('workspace_remove', { id: 'workspace-id' });

Backend command (src-tauri/src/lib.rs):

#[tauri::command]
async fn workspace_list(
    state: tauri::State<'_, AppState>,
) -> Result<Vec<Workspace>, String> {
    state.workspace_manager.lock().await.list_workspaces()
        .map_err(|e| e.to_string())
}

Thread Management

Thread Reducer Architecture

Thread state is managed by a reducer with slices in src/features/threads/hooks/threadReducer/.

Thread reducer pattern (src/features/threads/hooks/threadReducer/index.ts):

export type ThreadAction =
  | { type: 'SET_MESSAGES'; messages: Message[] }
  | { type: 'ADD_MESSAGE'; message: Message }
  | { type: 'UPDATE_MESSAGE'; messageId: string; updates: Partial<Message> }
  | { type: 'SET_RUNNING'; running: boolean }
  | { type: 'SET_UNREAD'; unread: number }
  | { type: 'RESET' };

export function threadReducer(state: ThreadState, action: ThreadAction): ThreadState {
  switch (action.type) {
    case 'SET_MESSAGES':
      return { ...state, messages: action.messages };
    case 'ADD_MESSAGE':
      return { ...state, messages: [...state.messages, action.message] };
    case 'UPDATE_MESSAGE':
      return {
        ...state,
        messages: state.messages.map(m =>
          m.id === action.messageId ? { ...m, ...action.updates } : m
        ),
      };
    case 'SET_RUNNING':
      return { ...state, running: action.running };
    case 'RESET':
      return initialThreadState;
    default:
      return state;
  }
}

Resuming a Thread

Frontend (src/features/threads/hooks/useThreadResume.ts):

import { invoke } from '@tauri-apps/api/tauri';

async function resumeThread(workspaceId: string, threadId: string) {
  const result = await invoke<{ messages: Message[] }>('thread_resume', {
    workspaceId,
    threadId,
  });
  
  dispatch({ type: 'SET_MESSAGES', messages: result.messages });
  dispatch({ type: 'SET_UNREAD', unread: 0 });
}

Backend (src-tauri/src/codex/mod.rs):

#[tauri::command]
async fn thread_resume(
    workspace_id: String,
    thread_id: String,
    state: tauri::State<'_, AppState>,
) -> Result<serde_json::Value, String> {
    let manager = state.workspace_manager.lock().await;
    let workspace = manager.get_workspace(&workspace_id)
        .ok_or("Workspace not found")?;
    
    let server = state.codex_servers.lock().await
        .get(&workspace_id)
        .ok_or("Server not running")?;
    
    server.call("thread/resume", json!({ "thread_id": thread_id })).await
        .map_err(|e| e.to_string())
}

Sending a Message

async function sendMessage(
  workspaceId: string,
  threadId: string,
  content: string,
  attachments?: { path: string; mime_type: string }[]
) {
  await invoke('thread_send_message', {
    workspaceId,
    threadId,
    message: {
      role: 'user',
      content,
      attachments,
    },
  });
}

Thread Lifecycle Commands

// Stop in-flight turn
await invoke('thread_interrupt', { workspaceId, threadId });

// Pin thread
await invoke('thread_pin', { workspaceId, threadId, pinned: true });

// Rename thread
await invoke('thread_rename', { workspaceId, threadId, name: 'New Name' });

// Archive thread
await invoke('thread_archive', { workspaceId, threadId });

// Copy thread (clone messages)
await invoke('thread_copy', { workspaceId, threadId });

Worktree Agents

Worktree agents create isolated git worktrees under <app-data>/worktrees/<workspace-id>/.

Creating a Worktree

Frontend:

const worktree = await invoke<{ path: string; branch: string }>('worktree_create', {
  workspaceId: 'workspace-id',
  branch: 'feature-branch',
});

console.log(`Worktree created at ${worktree.path}`);

Backend (src-tauri/src/shared/workspaces_core/worktree.rs):

pub async fn create_worktree(
    workspace_path: &str,
    workspace_id: &str,
    branch: &str,
    app_data_dir: &Path,
) -> Result<Worktree, WorktreeError> {
    let worktree_dir = app_data_dir.join("worktrees").join(workspace_id);
    std::fs::create_dir_all(&worktree_dir)?;
    
    let worktree_path = worktree_dir.join(branch);
    
    let output = Command::new("git")
        .args(&["worktree", "add", worktree_path.to_str().unwrap(), branch])
        .current_dir(workspace_path)
        .output()?;
    
    if !output.status.success() {
        return Err(WorktreeError::GitError(
            String::from_utf8_lossy(&output.stderr).to_string()
        ));
    }
    
    Ok(Worktree {
        path: worktree_path.to_string_lossy().to_string(),
        branch: branch.to_string(),
    })
}

Listing Worktrees

const worktrees = await invoke<Worktree[]>('worktree_list', {
  workspaceId: 'workspace-id',
});

Removing a Worktree

await invoke('worktree_remove', {
  workspaceId: 'workspace-id',
  path: '/path/to/worktree',
});

Git Integration

Git Diff Stats

Frontend:

const stats = await invoke<{
  staged: { path: string; status: string }[];
  unstaged: { path: string; status: string }[];
}>('git_diff_stats', { workspaceId: 'workspace-id' });

Backend (src-tauri/src/shared/git_ui_core/diff.rs):

pub fn get_diff_stats(repo_path: &str) -> Result<DiffStats, GitError> {
    let repo = Repository::open(repo_path)?;
    let mut index = repo.index()?;
    
    let head_tree = repo.head()?.peel_to_tree()?;
    let diff_index_tree = repo.diff_tree_to_index(Some(&head_tree), Some(&index), None)?;
    let diff_index_workdir = repo.diff_index_to_workdir(Some(&index), None)?;
    
    let staged = collect_diff_entries(&diff_index_tree)?;
    let unstaged = collect_diff_entries(&diff_index_workdir)?;
    
    Ok(DiffStats { staged, unstaged })
}

Branch Management

// List branches
const branches = await invoke<{ name: string; current: boolean; ahead: number; behind: number }[]>(
  'git_list_branches',
  { workspaceId: 'workspace-id' }
);

// Checkout branch
await invoke('git_checkout_branch', {
  workspaceId: 'workspace-id',
  branch: 'main',
});

// Create branch
await invoke('git_create_branch', {
  workspaceId: 'workspace-id',
  branch: 'feature-new',
  fromBranch: 'main',
});

GitHub Integration

Requires gh CLI:

// List issues
const issues = await invoke<GitHubIssue[]>('github_list_issues', {
  workspaceId: 'workspace-id',
});

// List PRs
const prs = await invoke<GitHubPR[]>('github_list_prs', {
  workspaceId: 'workspace-id',
});

// Get PR diff
const diff = await invoke<string>('github_pr_diff', {
  workspaceId: 'workspace-id',
  prNumber: 42,
});

// Ask PR (send PR context to new thread)
await invoke('github_ask_pr', {
  workspaceId: 'workspace-id',
  prNumber: 42,
  question: 'What does this PR change?',
});

Backend (src-tauri/src/shared/git_ui_core/github.rs):

pub async fn list_prs(repo_path: &str) -> Result<Vec<GitHubPR>, GitHubError> {
    let output = Command::new("gh")
        .args(&["pr", "list", "--json", "number,title,author,state"])
        .current_dir(repo_path)
        .output()
        .await?;
    
    if !output.status.success() {
        return Err(GitHubError::CliError(
            String::from_utf8_lossy(&output.stderr).to_string()
        ));
    }
    
    Ok(serde_json::from_slice(&output.stdout)?)
}

Remote Daemon Mode

Remote daemon mode allows running Codex on a separate machine (e.g., desktop) and connecting from iOS or other clients.

Desktop Daemon Setup

Via UI:

  1. Settings → Server
  2. Set Remote backend token (shared secret)
  3. Click Start daemon in Mobile access daemon

Standalone Daemon CLI:

# Build daemon binaries
cd src-tauri
cargo build --bin codex_monitor_daemon --bin codex_monitor_daemonctl

# Start daemon (reads settings.json from app data dir)
./target/debug/codex_monitor_daemonctl start

# Status
./target/debug/codex_monitor_daemonctl status

# Stop
./target/debug/codex_monitor_daemonctl stop

# Override settings
./target/debug/codex_monitor_daemonctl start \
  --listen 0.0.0.0:4732 \
  --token $REMOTE_TOKEN \
  --data-dir /path/to/app/data

Daemon RPC Architecture

Daemon entrypoint (src-tauri/src/bin/codex_monitor_daemon.rs):

#[tokio::main]
async fn main() -> Result<()> {
    let listener = TcpListener::bind(&args.listen).await?;
    let shared_state = Arc::new(DaemonState::new(app_data_dir, codex_path)?);
    
    loop {
        let (socket, _) = listener.accept().await?;
        let state = shared_state.clone();
        tokio::spawn(handle_connection(socket, state));
    }
}

async fn handle_connection(socket: TcpStream, state: Arc<DaemonState>) {
    let (reader, writer) = socket.into_split();
    let reader = BufReader::new(reader);
    let mut writer = BufWriter::new(writer);
    
    let mut lines = reader.lines();
    while let Some(line) = lines.next_line().await? {
        let request: JsonRpcRequest = serde_json::from_str(&line)?;
        let response = handle_rpc_request(request, &state).await;
        writer.write_all(serde_json::to_string(&response)?.as_bytes()).await?;
        writer.write_all(b"\n").await?;
        writer.flush().await?;
    }
}

RPC routing (src-tauri/src/bin/codex_monitor_daemon/rpc.rs):

pub async fn handle_rpc_request(
    request: JsonRpcRequest,
    state: &DaemonState,
) -> JsonRpcResponse {
    match request.method.as_str() {
        "workspace/list" => workspace_list(state).await,
        "workspace/add" => workspace_add(request.params, state).await,
        "thread/list" => thread_list(request.params, state).await,
        "thread/resume" => thread_resume(request.params, state).await,
        "thread/send" => thread_send(request.params, state).await,
        "git/diff_stats" => git_diff_stats(request.params, state).await,
        _ => JsonRpcResponse::error(-32601, "Method not found"),
    }
}

iOS Client Connection (Tailscale)

Desktop (Tailscale helper):

  1. Settings → Server → Tailscale helper
  2. Click Detect Tailscale → note suggested host (e.g., your-mac.your-tailnet.ts.net:4732)

iOS:

  1. Settings → Server
  2. Enter desktop Tailscale host and matching token
  3. Tap Connect & test

Frontend client (src/services/remoteDaemon.ts):

class RemoteDaemonClient {
  private socket: WebSocket | null = null;
  private requestId = 0;
  
  async connect(host: string, token: string): Promise<void> {
    this.socket = new WebSocket(`ws://${host}`);
    
    await new Promise((resolve, reject) => {
      this.socket!.onopen = () => {
        this.send('auth', { token }).then(resolve).catch(reject);
      };
      this.socket!.onerror = reject;
    });
  }
  
  async send(method: string, params: any): Promise<any> {
    const id = ++this.requestId;
    const request = { jsonrpc: '2.0', id, method, params };
    
    return new Promise((resolve, reject) => {
      const handler = (event: MessageEvent) => {
        const response = JSON.parse(event.data);
        if (response.id === id) {
          this.socket!.removeEventListener('message', handler);
          if (response.error) {
            reject(new Error(response.error.message));
          } else {
            resolve(response.result);
          }
        }
      };
      
      this.socket!.addEventListener('message', handler);
      this.socket!.send(JSON.stringify(request));
    });
  }
}

Composer & Autocomplete

Autocomplete Triggers

  • $ — skills
  • /prompts: — prompts
  • /review — code review
  • @ — file paths

Frontend autocomplete hook (src/features/composer/hooks/useAutocomplete.ts):

export function useAutocomplete(value: string, cursorPosition: number) {
  const [suggestions, setSuggestions] = useState<Suggestion[]>([]);
  
  useEffect(() => {
    const prefix = value.slice(0, cursorPosition);
    
    if (prefix.endsWith('$')) {
      // Fetch skills
      invoke<string[]>('autocomplete_skills', { prefix }).then(setSuggestions);
    } else if (prefix.includes('/prompts:')) {
      // Fetch prompts
      invoke<Prompt[]>('autocomplete_prompts', { prefix }).then(setSuggestions);
    } else if (prefix.endsWith('@')) {
      // Fetch file paths
      invoke<string[]>('autocomplete_files', {
        workspaceId,
        prefix,
      }).then(setSuggestions);
    } else {
      setSuggestions([]);
    }
  }, [value, cursorPosition]);
  
  return suggestions;
}

Follow-up Behavior

Settings → Composer → Follow-up default:

  • Queue — queue messages if agent is running
  • Steer — interrupt current turn and steer

Override for single message:

  • macOS: Shift+Cmd+Enter
  • Windows/Linux: Shift+Ctrl+Enter

Prompt Library

Prompts load from $CODEX_HOME/prompts (or ~/.codex/prompts).

Prompt File Format

---
description: Generate unit tests for a function
args:
  - name: function_name
    description: Name of the function to test
---

Generate comprehensive unit tests for the function `{{function_name}}`, covering edge cases and error handling.

Managing Prompts

// List prompts
const prompts = await invoke<Prompt[]>('prompt_list', { workspaceId });

// Create prompt
await invoke('prompt_create', {
  workspaceId,
  name: 'generate-tests',
  content: '...',
  isGlobal: false, // workspace-specific
});

// Run prompt in current thread
await invoke('prompt_run', {
  workspaceId,
  threadId,
  promptId: 'generate-tests',
  args: { function_name: 'calculateTotal' },
});

// Run prompt in new thread
await invoke('prompt_run_new_thread', {
  workspaceId,
  promptId: 'generate-tests',
  args: { function_name: 'calculateTotal' },
});

Configuration

App Settings

Persisted to settings.json in app data directory.

interface AppSettings {
  theme: 'light' | 'dark' | 'system';
  backend_mode: 'local' | 'remote';
  remote_provider?: 'tcp' | 'ws';
  remote_tcp_host?: string;
  remote_tcp_token?: string;
  codex_path?: string; // Custom Codex binary path
  default_access_mode?: 'default' | 'direct' | 'tool_only';
  ui_scale?: number;
  follow_up_behavior?: 'queue' | 'steer';
  reduced_transparency?: boolean;
}

// Get settings
const settings = await invoke<AppSettings>('get_app_settings');

// Update settings
await invoke('update_app_settings', {
  updates: { theme: 'dark', ui_scale: 1.2 },
});

Codex Config

Feature settings sync to $CODEX_HOME/config.toml:

[features]
collaboration_modes = true
unified_exec = true  # Background terminal
apps = false  # Experimental

[personality]
tone = "professional"

Load/save via:

const config = await invoke<CodexConfig>('get_codex_config_path');

await invoke('codex_doctor'); // Validate Codex setup

iOS Development

Build for Simulator

./scripts/build_run_ios.sh
# Options: --simulator "<name>", --target aarch64-sim|x86_64-sim, --skip-build, --no-clean

Build for USB Device

# List devices
./scripts/build_run_ios_device.sh --list-devices

# Build and run
./scripts/build_run_ios_device.sh --device "<device name>" --team <TEAM_ID>
# Options: --target aarch64, --skip-build, --bundle-id <id>

Signing Configuration

Preferred: src-tauri/tauri.ios.local.conf.json (gitignored):

{
  "bundle": {
    "iOS": {
      "developmentTeam": "YOUR_TEAM_ID"
    }
  },
  "identifier": "com.yourcompany.codexmonitor"
}

TestFlight Release

# Copy .testflight.local.env.example to .testflight.local.env and fill values
./scripts/release_testflight_ios.sh

Required env vars in .testflight.local.env:

IOS_TEAM_ID=YOUR_TEAM_ID
BUNDLE_ID=com.yourcompany.codexmonitor
TESTFLIGHT_BETA_GROUP="Beta Testers"
APPLE_ID=your-apple-id@example.com
APP_STORE_CONNECT_TEAM_ID=YOUR_ASC_TEAM_ID

Terminal Dock

Experimental feature for background commands.

// Execute command in terminal tab
await invoke('terminal_exec', {
  workspaceId,
  tabId: 'tab-1',
  command: 'npm test',
});

// Create new terminal tab
await invoke('terminal_create_tab', { workspaceId, name: 'Tests' });

// Close terminal tab
await invoke('terminal_close_tab', { workspaceId, tabId: 'tab-1' });

Troubleshooting

Native Build Errors

npm run doctor

Common issues:

  • CMake not found: Install CMake
  • bindgen errors (Windows): Install LLVM/Clang
  • Rust targets missing (iOS): rustup target add aarch64-apple-ios aarch64-apple-ios-sim

Codex Not Found

Set custom Codex path in Settings → General → Codex binary path, or ensure codex is in PATH.

Remote Daemon Connection Fails

  1. Confirm daemon is running: ./target/debug/codex_monitor_daemonctl status
  2. Check token matches between desktop and client
  3. Verify host/port reachable (Tailscale: both devices online in same tailnet)
  4. Check firewall rules for listening port (default 4732)

Thread Resume Shows Stale Messages

Threads are restored from disk via thread/resume. If messages are stale:

  1. Ensure workspace cwd matches thread working directory
  2. Restart workspace server: Remove and re-add workspace

Worktree Creation Fails

Ensure workspace is a git repository:

cd /path/to/workspace
git status

Legacy worktrees under .codex-worktrees/ are supported but new ones use <app-data>/worktrees/<workspace-id>/.

iOS Signing Issues

First-time device setup:

  1. iPhone unlocked and trusted with Mac
  2. Developer Mode enabled on iPhone (Settings → Privacy & Security → Developer Mode)
  3. Open Xcode via ./scripts/build_run_ios_device.sh --open-xcode and approve signing

File Structure Reference

src/
  features/app/bootstrap/         App bootstrap orchestration
  features/app/orchestration/     Layout/thread/workspace orchestration
  features/threads/hooks/threadReducer/  Thread reducer slices
  features/composer/              Composer UI and autocomplete
  features/git/                   Git UI components
  features/prompts/               Prompt library UI
  services/tauri.ts               Tauri IPC wrapper
  types.ts                        Shared TypeScript types

src-tauri/
  src/lib.rs                      Tauri command registry
  src/bin/codex_monitor_daemon.rs Remote daemon entrypoint
  src/bin/codex_monitor_daemon/rpc/ Daemon RPC handlers
  src/shared/git_ui_core/         Git/GitHub core
  src/shared/workspaces_core/     Workspace/worktree core
  src/workspaces/                 Workspace adapters
  src/codex/                      Codex app-server adapters
  src/files/                      File adapters

Resources

FAQ

What does the codexmonitor-orchestration skill do?

Expert in CodexMonitor, a Tauri app for orchestrating multiple Codex agents across local workspaces with threads, git integration, and remote daemon support.

How do I install codexmonitor-orchestration?

Run npx skills add https://github.com/reason-machines/codex-skills --skill codexmonitor-orchestration in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/codexmonitor-orchestration/ (personal) or .claude/skills/codexmonitor-orchestration/ (project).

Which AI agents can use codexmonitor-orchestration?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is codexmonitor-orchestration safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains codexmonitor-orchestration?

It is published by reason-machines in reason-machines/codex-skills. It has 265 recorded installs.

More from reason-machines/codex-skills

View all →

academic-research-skills-codex

reason-machines/codex-skills

Open the skill to read its SKILL.md instructions.

Productivity & Automation412 installs

codexplusplus-launcher

reason-machines/codex-skills

Open the skill to read its SKILL.md instructions.

282 installs

codex-plugin-claude-code

reason-machines/codex-skills

Open the skill to read its SKILL.md instructions.

273 installs

awesome-codex-skills-curator

reason-machines/codex-skills

Open the skill to read its SKILL.md instructions.

270 installs

codex-plusplus-tweak-system

reason-machines/codex-skills

Open the skill to read its SKILL.md instructions.

270 installs

awesome-codex-subagents

reason-machines/codex-skills

Open the skill to read its SKILL.md instructions.

AI & Agents268 installs

Popular code quality & workflow skills

View category →

grill-me

mattpocock/skills

A relentless interview to sharpen a plan or design.

Code Quality & Workflow1.4M installs

improve-codebase-architecture

mattpocock/skills

Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.

Code Quality & Workflow1.2M installs

handoff

mattpocock/skills

Compact the current conversation into a handoff document for another agent to pick up.

Code Quality & Workflow1.1M installs

domain-modeling

mattpocock/skills

Build and sharpen a project's domain model. Use when discussing codebase terminology, writing or editing a GLOSSARY.md, or recording or editing an ADR.

Code Quality & Workflow890.6K installs

codebase-design

mattpocock/skills

Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.

Code Quality & Workflow863.8K installs

diagnosing-bugs

mattpocock/skills

Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.

Code Quality & Workflow848.4K installs