Agenttrail CLI
A coding-agent monitor with a project component map and an experimental 3D kitchen view of ongoing tasks
- 980 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Task titles and paths can reveal project information
- Optional hook setup changes agent configuration
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/agenttrailDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Task titles and paths can reveal project information
- Optional hook setup changes agent configuration
This is third-party code. Review the repository files before installing.
What it does
Agenttrail displays coding-agent activity in a browser. It does not launch agents; they continue running in their own clients. The project has two views with different data sources. Map builds a component view from `PLAN.md` and shows changes to associated files. Optional Claude Code hooks add session and tool-call information. This lets you compare the project plan with the files being changed. Kitchen represents Codex and Claude Code tasks as orders in a 3D kitchen. Chefs represent roles, so one session can use several chefs in sequence. Basic observation needs accessible local logs; additional links between sessions are configured separately.
Who it is for. For developers running several Codex or Claude Code sessions who want a view of current tasks beside their editor.
Good fit when
- You want to see which project components change during agent work
- Several local sessions are active and their task lists are hard to follow in terminals
- You want a screen recording that makes project responsibilities visible
- You want to distinguish reported task completion from observed file changes
Not a fit when
- You need a coordinator that assigns tasks and launches agents
- You need exact model billing and a complete trace of every API request
- Sessions run in the cloud and their logs are unavailable on this machine
Example request
Open Kitchen for this repository and check whether it sees the current local Codex session. Leave project settings unchanged for now.Limitations
Kitchen requires Node.js 20 or later and a WebGL-capable browser. The release is experimental: Windows and live Cursor sessions are not yet validated, so start with a test project. A delivered order reflects task status and does not establish that tests passed or results were published. Map does not yet read the newer Claude Code TaskCreate and TaskUpdate lists, so its history can differ from Kitchen even when observing the same work.
How to disable. Stop the watcher in its terminal. If you enabled optional hooks, review the relevant agent settings and remove the added entries. Map stores history in `~/.agenttrail`; Kitchen keeps project selections and connector registrations in `~/.agent-office`. Inspect needed data before deleting either directory.
FAQ
How do I start Agenttrail Kitchen?
Run `npx agenttrail-kitchen .` in the project directory. The browser opens at `http://localhost:4780` or the next free port. Keep the terminal running.
Does Kitchen need a project plan?
No. Basic observation uses available local Codex or Claude Code logs. `npx agenttrail init` configures Map and writes project files.
Why are tasks missing?
Check the repository selection, Live mode and Connect agents. Logs must be locally accessible; discovering activity can take around five seconds.
Does Agenttrail upload logs to the cloud?
No. Agenttrail’s own server processes logs locally and binds to `127.0.0.1`. The browser displays paths and tasks, and Map may show shortened commands. The coding agent transfers data according to its own settings.
Can I preview it without an agent?
Yes. Run `npx agenttrail-kitchen . --example` and advance the example steps. The interface labels this scripted demonstration separately from live activity.
Does it need another model key?
No. Code and original assets use MIT. Changes to log formats can require an Agenttrail update to restore event reading.
Discussion
No questions or reviews yet.
Related
CLI toolsagentplay turns local Claude Code, Codex and Cursor sessions into a file map and a replayable step timeline.
One desktop window for Claude Code, Codex, Gemini CLI and OpenCode: terminals, pull requests, diffs and approvals
A CLI that builds a map of your repo's code and keeps team decisions in .trail, so Claude Code, Cursor and Codex stop re-exploring the project
A Claude Code and Codex plugin with skills for code research, tests, refactoring and independent review of finished changes
A Claude Code plugin links planning, implementation and task checks, retains progress and controls how independently agents work.
Over twenty skills for coding agents: MCP and hooks, plan mode, a bug-fix protocol, prompting and a bash safety hook
README in short
Map and Kitchen have separate services and local storage. Kitchen opens in a browser beside the editor; no dedicated VS Code extension is available yet.