alethe-agents

Name: alethe-agents
Author: Kc1t

Warning

A local-first desktop workspace for running, organizing, and resuming multiple coding agents and shells with real PTYs, split panes, persistent layouts, and local session history.

787stars

103forks

TypeScript

Installation

# Add to your Claude Code skills
git clone https://github.com/Kc1t/alethe-agents

Getting Started

Guides for using ai agents skills like alethe-agents.

Caveman: Cut Claude Token Use by 65%
How agent-side prompt compression works, when to use it, and when not to.
What is an AI Skills Marketplace?
Definitions, how marketplaces work, and how to choose between them in 2026.
Getting Started with AI Skills

Security ReportWarning

Last scanned: 8/19/2026

{
  "issues": [
    {
      "type": "npm-audit",
      "message": "dompurify: DOMPurify: IN_PLACE hook removal leaves a detached subtree executable, causing XSS",
      "severity": "medium"
    },
    {
      "type": "npm-audit",
      "message": "mermaid: Mermaid configuration APIs allow prototype pollution",
      "severity": "medium"
    },
    {
      "type": "npm-audit",
      "message": "nanoid: nanoid: non-secure generators can loop indefinitely with negative size",
      "severity": "high"
    }
  ],
  "status": "WARNING",
  "scannedAt": "2026-08-19T04:35:53.771Z",
  "npmAuditRan": true,
  "pipAuditRan": true,
  "promptInjectionRan": true
}

Compare with similar skills

See how alethe-agents compares with popular alternatives.

alethe-agents vs hermes agent

See comparison

alethe-agents vs ecc

See comparison

alethe-agents vs superpowers

See comparison

README.md

Frequently Asked Questions

What is alethe-agents?

alethe-agents is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by Kc1t. A local-first desktop workspace for running, organizing, and resuming multiple coding agents and shells with real PTYs, split panes, persistent layouts, and local session history. It has 787 GitHub stars.

Is alethe-agents safe to use?

alethe-agents returned warnings in SkillsLLM's automated security scan. It has no critical vulnerabilities, but review the flagged issues in the Security Report section before adding it to your workflow.

How do I install alethe-agents?

Clone the repository with "git clone https://github.com/Kc1t/alethe-agents" and add it to your Claude Code skills directory (see the Installation section above).

What programming language is alethe-agents written in?

alethe-agents is primarily written in TypeScript. It is open-source under Kc1t on GitHub, so you can review or fork the full source.

Are there alternatives to alethe-agents?

Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh alethe-agents against similar tools.

Agentic AI for Beginners

Build your first AI agent from scratch - tool use, ReAct pattern, memory, deployment

41 minBeginner

Comments (0)

to leave a comment.

No comments yet. Be the first to share your thoughts!

Related Skills

ECC

by affaan-m

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

270,878

doop agent-console

[!IMPORTANT] Alethe is an early public release. The desktop app is free, open source, and local-first, not local-only: update checks and provider usage polling are on by default, while other network features are optional or action-triggered. Manual GitHub Gist Sync is already available; first-party hosted sync or cloud backup may be offered separately later. See the privacy and data-flow guide.

What Alethe Is

One agent in one terminal is easy. Five agents across three repositories is the actual job — and that is where terminal tabs stop working: sessions get lost, MCP servers drift out of sync between agents, and nobody knows which agent is doing what, where.

Alethe is a desktop workspace built for that. Every agent runs in a real PTY inside a persistent project layout, keeps its own session and history, and stays alive when you rearrange the UI. On top of that, Alethe manages the things agents share: their CLIs, their MCP servers, their skills, and the conversations you move between them.

Cross-platform (Windows, macOS, Linux), local-first, built with Tauri, Rust, React, and xterm.js. “Local-first” describes workspace persistence, not an internet-free guarantee; see docs/PRIVACY.md for current network defaults, credentials, and retention.

Supported Platforms

Agents

Agent	CLI
Claude Code	`claude`	Session resume, usage cards, local history
Codex	`codex`	Session resume, usage cards
GitHub Copilot CLI	`copilot`
Cursor	`cursor-agent`	Session resume
Antigravity	`agy`	Usage cards
OpenCode	`opencode`	Session resume
Kiro CLI	`kiro-cli`
Kimi Code	`kimi`
Grok Build	`grok`
Codewhale	`codewhale`
Mimo	`mimo`
Freebuff	`freebuff`
Shell	pwsh / bash / zsh	The plain terminal, same pane model
WSL	`wsl.exe`	The default distro, as a plain shell (Windows)

Missing CLIs can be installed, updated, and uninstalled from inside Alethe — it probes the machine for Node, npm, WinGet, Scoop, and Chocolatey and offers only the methods that actually work there, preferring each vendor's official installer. Already-installed CLIs are discovered across PATH, registry, npm/pnpm/Volta/fnm/nvm/Bun/Cargo/Scoop/Chocolatey, and can be pointed at a custom path. A plugin can add an agent of its own to this list.

Agents can optionally be routed through 9router, a local proxy that spreads traffic across providers with automatic fallback. Alethe installs and runs a private, version-pinned copy without touching your global npm packages. Routing is off by default and applies only to terminals opened after it is switched on.

What It Does

Run agents in parallel

Projects, groups, and subgroups organize repositories; each open project becomes a container with its own panes.
One agent per pane, or several agents as sub-tabs inside the same pane — each with its own PTY, working directory, and session.
Auto, spotlight, sidebar, and custom grid layouts, editable directly on the grid.
Closing a container hides it; the process keeps running.

Keep the context

Sessions of Claude Code, Codex, Cursor, and OpenCode resume after a crash or a restart.
Recent chats lists the conversations of a pane's working directory and reopens any of them.
A Claude Code conversation can be handed off to Codex (and back) through a locally redacted context packet — no copy-pasting the thread by hand. Redaction is best effort, so review the packet before starting the target agent.
Scrollback is persisted per PTY, so reattaching shows what happened before.

Manage what the agents share

MCP tab: every MCP server configured on the machine, grouped by server and showing which agents have it — read from Claude Code, Codex, Cursor, OpenCode, and Antigravity configs. Add, remove, copy a server from one agent to another, search the official registry, and ask each agent to verify it can really reach a server. Every write is backed up, re-parsed, and committed atomically.
Skills tab: the skills installed for each agent, with links and shared stores resolved so a shared skill shows up once.
Graphify: a code graph of the project, served to the agents as an MCP server.

Extend it with plugins

Features load as plugins, not hard-wired code. Official ones — Todo List, Git Control, Theme Pack — ship inside the installer and can be switched off in Preferences ▸ Plugins, with their surfaces appearing and disappearing without a restart.
A plugin can contribute a workspace pane, a sidebar tab, a command palette entry, a full theme, or a new agent provider. It declares what it needs in its manifest, and the app refuses anything it did not ask for.
Plugin catalogue: plugins published by other people are listed inside Alethe with their permissions spelled out, and a listing that ships a package installs — and updates — in one click. The index pins every package to a SHA-256 checksum, so what runs is what the listing was reviewed against. An installed plugin arrives switched off and goes through a trust dialog before it runs.

Publishing a plugin

A plugin is a folder whose name matches the id in its plugin.json, next to a main.js bundle that binds against window.alethe (React included — never bundle your own). The manifest declares the tabs, commands and capabilities; the code implements them.

Try it locally. Drop the folder into <profile>/plugins/<id>/, or use Preferences ▸ Plugins ▸ Import plugin. It arrives switched off — enabling it goes through the trust dialog.
Ship a zip. Zip the plugin folder, take its SHA-256, and attach it to a release: zip -r my-plugin.zip my-plugin && sha256sum my-plugin.zip.
Get it listed. Open a pull request against this repository adding one entry to plugins.json with id, name, downloadUrl and — to make it installable from inside Alethe — package.url and package.sha256. Re-hash whenever the zip changes.
Once merged, the listing reaches everyone within six hours, or immediately on Refresh.

Listing is a human reading your pull request, not an audit: the plugin still arrives disabled and still goes through the trust dialog. docs/examples/notes-plugin/ is a working plugin in plain JavaScript with no build step; the plugin guide has the full contract.

Stay in control

RAM readout in the title bar; disable a terminal or suspend a whole group to get memory back.
Git panel per project — status, stage, commit, branches, diffs in a pane — plus worktrees for parallel tasks.
Pull Request review: the merge panel finds the open GitHub PR for an agent worktree through the local gh CLI, opens its metadata, and can start an AI review inside the same isolated worktree — read-only, with no commits, pushes, merges, or GitHub comments. Squash-merging stays an explicit human actio