Chops
The macOS app to discover, organize, and edit coding agent skills across Claude Code, Cursor, Codex, Windsurf, and Amp.
- Source repo
- Shpigford/chops
- Stars
- ★ 1.9k
- Last updated
- 1mo ago
- License
- NOASSERTION
- Primary language
- Swift
- FA score
- 39/100 · Major gaps
At a glance
- How it runs
- Works with
- Platform-specific
- Cost
- Free, no paid service needed
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- Developers juggling skills across Claude Code and Cursor who want a single interface to browse and edit them.
- Not a fit if
- Windows or Linux users: macOS 15+ only
- Users unwilling to grant full filesystem access: sandbox is intentionally disabled
- Teams needing automated test coverage: no test suite exists
- Source review
- 39/100 · Major gaps 4 safety controls not found
What does this agent do, and when should you use it?
Chops is a native macOS app (SwiftUI + SwiftData) that centralizes management of AI coding skills and agents from multiple tools. It automatically scans tool directories (Claude Code, Cursor, Codex, etc.), parses, deduplicates, and watches files in real time, with a built-in editor, full-text search, and collections. The app runs unsandboxed for necessary file access. The repository includes a complete Xcode project configuration (project.yml), build scripts, a marketing site, and a setup skill for AI coding agents. It requires macOS 15+ and offers manual build via Xcode.
Chops uses SkillScanner to probe each tool's skill and agent directories (e.g., ~/.claude/skills) and parses YAML frontmatter from Markdown or Cursor .mdc files, storing skill records in SwiftData. FileWatcher uses FSEvents to monitor disk changes and triggers rescanning automatically. The app provides a three-column NavigationSplitView: a sidebar with tool filters and collections, a list of filtered/searched skills, and a detail editor wrapping NSTextView with Cmd+S save and frontmatter parsing. It supports creating new skills/agents with tool-specific boilerplate. Skills are deduplicated via resolved symlink paths, showing multiple tool badges when shared. Remote servers running OpenClaw, Hermes, or similar layouts can be connected to discover, browse, and install skills.
- Developers juggling skills across Claude Code and Cursor who want a single interface to browse and edit them.
- Teams that need to organize shared agents into collections without altering source files.
- Users who want to quickly search for specific content within all their skills across tools.
- Users creating new skills or agents for multiple tools, avoiding manual template writing.
- Administrators who need real-time monitoring of skill file changes and instant updates.
- Researchers or ops teams connecting to remote servers (like OpenClaw or Hermes) to discover and install skills.
How do you install or deploy this agent?
Download Chops.dmg from the GitHub Releases page. To build from source: git clone https://github.com/Shpigford/chops.git, cd chops, install xcodegen (brew install xcodegen), run xcodegen generate to create the Xcode project, then open Chops.xcodeproj and press Cmd+R to run.
How do you use this agent?
After launching, the app scans supported skill directories automatically. Use the sidebar to filter by tool or collection, select a skill from the list, edit in the detail pane with the built-in editor, and save with Cmd+S. Create new skills/agents via the UI. Use the search box for full-text search. To connect to a remote server, use settings to add the server URL (for OpenClaw or Hermes).
What are this agent's strengths and limitations?
- Supports multiple tools (Claude Code, Cursor, Codex, Windsurf, Amp) with unified management.
- Built-in editor with frontmatter parsing and Cmd+S save.
- Real-time file watching via FSEvents keeps data fresh.
- Full-text search, collections, and create/new skills with boilerplate generation.
- Clear architecture; adding a new tool is straightforward through the ToolSource enum.
- macOS-only (15+), no Windows/Linux support.
- Requires local build with Xcode and xcodegen; distribution relies on GitHub Releases.
- Unsandboxed app requires filesystem access, raising security concerns for some users.
- No automated test suite; manual validation required, increasing maintenance risk.
- Remote server support limited to specific protocols/layouts (OpenClaw, Hermes).
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Chops This agent | 39 · Major gaps | Desktop appFree | ★ 1.9k | 1mo ago | Swift | — |
| Agent Sessions | 75 · Good | Desktop appFree | ★ 905 | today | Swift | Codex · Claude Code |
| Coucou Agent Companion | 73 · Some gaps | Desktop appFree + model costs | ★ 5.5k | today | Swift | Codex · Claude Code |
| CoCo Super Intelligence | 69 · Some gaps | Agent plugin / skillFree + model costs | ★ 744 | today | Python | Codex · Claude Code |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows the app explicitly disables sandbox (Chops.entitlements), requiring unrestricted filesystem access to read dotfiles in user home directories, which is necessary for core functionality but grants broad permissions without user confirmation mechanisms. Data flow transparency is insufficient; no details on how skill files are read, stored, or transmitted. Sensitive data handling is not mentioned. Dependency security: only Sparkle is mentioned as external dependency, but no lock files or vulnerability scanning evidence. External effects include file watching and auto-update, but user control details are lacking. Rollback mechanisms are absent. Source attribution: copyright to Sabotage Media LLC, but publisher is unverified.
Self-consistency is good; README architecture matches project structure, but no test suite, so reliability evidence is limited. Dependency availability: relies on Xcode, Homebrew, xcodegen, but no version pinning. Failure messages: README mentions manual validation of error messages, but no concrete examples.
Target audience is clear (macOS developers), use cases are well-defined. Capability boundaries are described in README, but unsupported features are not explicitly listed. Trigger precision: file watching and scanning are automatic, but user configuration details are missing. Environment fit: explicitly requires macOS 15 and Xcode, but no mention of other OS support.
Information architecture is clear; project structure is documented. Install notes are detailed, including Quick Start and CLI build. Naming stability: tool names and paths are listed, but no version history. Examples and FAQ are sparse; only one AI Agent Setup example. Known limitations are not explicitly listed. License is FSL-1.1-MIT, but metadata is NOASSERTION. Versioning and changelog are not mentioned. Maintenance responsibility: publisher unverified, but README provides contact info.
Output usability: app provides editor, search, and collections, but no detailed output format. Marginal value: addresses skill management pain point, but no comparison with other tools. Cost-benefit: requires macOS 15 and Xcode, which may increase adoption cost, but no performance or resource usage data.
Claim traceability: README feature descriptions align with code structure, but no implementation details. Cross-source corroboration: relies solely on single source, no external validation. Fact-inference separation: design decisions (e.g., no sandbox) are explicit, but facts and inferences are not clearly distinguished.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- Not found in source: data-flow disclosureWatch which external services it contacts (proxy or firewall logs) and keep sensitive data out until you know where it goes.
- Not found in source: sensitive-data handlingUse dedicated, low-privilege, revocable API keys — never production credentials — and keep secrets out of logs.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- The app disables sandbox, granting broad permissions; evaluate carefully.
- No user confirmation mechanisms; automatic scanning and updates may affect user control.
- Data flow transparency is insufficient; clarify how skill files are handled.
- Dependency security evidence is lacking; consider checking dependency locking and vulnerability scanning.
- Publisher is unverified; be cautious about source credibility.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
v1.16.0 Chops v1.16.0
Release notes excerpt
- Browse Skills now opens to a Trending list of popular skills - Faster, broader skill search with an Official-only filter - Agents and rules now live in one unified Skills list, with a Discovery row to browse the registry - New ⌘K command palette to jump to any tool's library, browse the registry, or create a skill/agent/rule - Added keyboard shortcuts: ⌘B to toggle the sidebar and ⌘⇧L to jump to SkillsView on GitHub ↗ -
v1.15.0 Chops v1.15.0
Release notes excerpt
- AI Assist now drives your installed Claude and Codex CLIs directly — fewer moving parts, more reliable responsesView on GitHub ↗ -
v1.14.0 Chops v1.14.0
Release notes excerpt
- Add Hermes as a tool source - Fixed editor diffs overwriting content that hadn't changed - Added Opus model option to the Claude agent model pickerView on GitHub ↗