Skip to content
shamclerenPublic

About

Unified monitoring panel for multiple IDEs and AI agents

Resources

Stars

17 stars

Watchers

1 watching

Forks

Latest commit

 

History

188 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CodePal icon CodePal

A local AI coding control tower for heavy agent users — sessions, usage, work review, and diagnostics in one floating macOS panel.

platform macOS version license MIT GitHub stars
Download from Releases · Product page · 简体中文


Why CodePal

Running multiple AI coding agents means your attention is constantly split:

  • one session is running in Cursor while another waits in a terminal
  • token and cost signals are scattered across model providers and local logs
  • finished work is buried inside transcripts instead of becoming a daily review
  • integration health is spread across hooks, config files, terminals, and desktop clients

CodePal pulls that into one local panel that stays visible while you work, then turns the observed history into usage analytics and daily operations memory.

Non-Intrusive by Design

CodePal is a dashboard, not a middleman. The agents you already use stay in charge of approval, execution, and output — CodePal just watches.

  • Native flow untouched. Approval prompts, tool calls, and decisions still happen inside Claude Code / Cursor / Codex / CodeBuddy. CodePal never blocks, gates, or rewrites them.
  • Visibility-only default. If CodePal is closed, crashing, or mid-update, your session keeps running exactly as if CodePal weren't installed. Every integration is additive.
  • Gracefully degradable. All hooks fail open. The worst case is a missed status update in the dashboard — never a stuck agent.

Preview

CodePal Dashboard

Watch the scripted walkthrough video · Open the static product page

Usage Analytics Daily Work Review
CodePal Analytics CodePal Work Review
Local Provider Gateway
CodePal Provider Gateway

What You Get

  • Unified session view: active, waiting, completed, errored, and restored sessions across supported agents in one list.
  • Focused activity timeline: replies, tool calls, status changes, and low-noise assistant/tool markers without transcript clutter taking over the row.
  • Per-session usage footer: requests, input, output, cache, context pressure, and estimated cost on expanded sessions.
  • Token usage analytics: model, agent, project, token-type, cache, cost, and trend views, backed by local history and HTML reports.
  • Daily Work Review: recent sessions grouped by day and project so finished, running, and waiting work can be reviewed quickly.
  • Work health and Attention signals: waiting sessions, context pressure, unrecovered failures, and follow-up needs rise to the top.
  • Session history persistence: normalized activity history stored locally and restored across app restarts.
  • Local Provider Gateway: connect supported desktop clients to third-party providers while keeping real provider tokens inside CodePal.
  • Capability-gated local operations: notifications, click-to-navigate, terminal message delivery, integration repair, and local action surfaces stay user-triggered.
  • Bilingual UI: English and Simplified Chinese, following your system language by default.

Supported Agents

Agent Session Usage
Cursor ✅ ✅
Claude Code ✅ ✅
Codex ✅ ✅
CodeBuddy ✅ ✅
GoLand / PyCharm* ✅ ✅

* GoLand and PyCharm flow through the shared CodeBuddy JetBrains plugin path.

Privacy Boundary

CodePal is local-first and monitoring-first:

  • it reads local agent session/transcript logs only for supported integrations
  • it does not upload prompts, transcripts, or repository contents to a CodePal cloud backend
  • it does not become an approval interceptor or autonomous scheduler
  • outbound actions are bounded, capability-gated, explicitly user-triggered, and limited to CodePal-owned sessions

See Privacy and Data Boundaries for details.

Install

  1. Open Releases.
  2. Download the latest .dmg or .zip for macOS.
  3. Move CodePal.app into Applications.
  4. Launch — connected agents are picked up automatically.

Release builds are signed and notarized by Apple. No security prompt on open.

What's Next

  • ACP Sessions: use Agent Client Protocol as the future operation entry for sessions CodePal creates or explicitly owns, instead of injecting text or decisions into native agent sessions.
  • Work item flow: turn observed sessions into actionable handoff, follow-up, failure, and completion records without taking over the original agent UI.
  • Manual LLM reports when useful: daily, weekly, and monthly reports from deterministic local facts, with redaction controls before anything leaves the app.
  • Free local control tower first: make the individual workflow strong enough for daily use before revisiting shared ops visibility, billing, or cloud sync.

See docs/planning/roadmap-next.md for the full planning direction.

Quick Start (Development)

git clone https://github.com/shamcleren/CodePal.git
cd CodePal
npm install
npm run dev        # launch in dev mode
npm run test       # run unit tests
npm run dist:mac   # build .dmg / .zip (requires Apple signing credentials)

To produce a signed and notarized build, set APPLE_ID, APPLE_APP_SPECIFIC_PASSWORD, and APPLE_TEAM_ID before running dist:mac.

Troubleshooting

Sessions not showing up Make sure the corresponding agent (Cursor / Claude Code / Codex / CodeBuddy) is actually running a session. Use the in-app diagnostics page to verify the integration path is healthy.

Privacy and Support

Development

Internal docs

License

MIT

About

Unified monitoring panel for multiple IDEs and AI agents

Resources

Stars

17 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages