agentic-stack CLI
A portable .agent folder with memory, skills and protocols that plugs into many coding agents and keeps its knowledge when you switch tools
- 2,309 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Loop mode runs agents that change code in worktrees
- Installs hooks and scripts into the project
Install
- Works in
Manual install
brew tap codejunkie99/agentic-stack https://github.com/codejunkie99/agentic-stack
brew install agentic-stack
cd your-project
agentic-stack claude-codeReplace claude-code with the adapter you need: cursor, windsurf, opencode, codex, gemini and others.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Loop mode runs agents that change code in worktrees
- Installs hooks and scripts into the project
This is third-party code. Review the repository files before installing.
What it does
The installer drops a .agent folder with lesson memory, skills and protocols into the project and wires it to chosen agents through adapters: Claude Code, Cursor, Windsurf, OpenCode, Codex, Gemini CLI and others. The dashboard, status and doctor commands show which adapters are wired and their health. Loop mode runs bounded agent loops in owned git worktrees with attempt, time and token budgets and an independent checker. A local data layer tracks agent activity, with optional integration with the external Brain memory.
Who it is for. For developers who use several coding agents and want shared memory and skills across them.
Good fit when
- Claude Code, Cursor or other agents share a project and their rules drift apart
- Lessons from past sessions must survive switching agents
- You need budget-bounded agent loops, for example to fix failing tests
Not a fit when
- You use one agent and its built-in memory is enough
- Third-party scripts and hooks are not allowed in the project
Example request
Set up agentic-stack in this project for Claude Code and Cursor and check adapter healthLimitations
Adds many files and Python scripts to the project. The loop supervisor bounds child processes but is not an OS sandbox; use the agent's own sandbox for isolation. The remove command deletes an adapter with no undo.
How to disable. Run ./install.sh remove <adapter> for each wired agent, or delete the .agent folder and adapter files manually.
FAQ
Can several agents share one project?
Yes, the add command wires a second adapter, for example ./install.sh add cursor.
How do I update skills and infrastructure?
Preview with ./install.sh upgrade --dry-run, then run ./install.sh upgrade --yes.
Discussion
No questions or reviews yet.
Related
CLI toolsA memory and coordination layer for coding agents: specs and issues live in git while agents work in parallel
A tmux-based TUI that runs Claude Code, Codex and other agents in one list: status, quick prompts, worktrees and diff review
Local persistent memory for coding agents: hooks record every session, and Claude Code, Codex or Cursor recall past decisions through 54 MCP tools
A local dashboard for running many agents across machines: memory, code graph and agent-to-agent messaging
Memory for coding agents in plain Markdown: the agent keeps specs, decisions and lessons itself, and each new session starts from them. No vector database
A CLI that keeps coding agents' shared context in one place: instructions, skills, MCPs, subagents and durable memory across projects and machines
README in short
The README explains one memory-and-skills folder for many agents and lists supported adapters. Install via Homebrew, or clone and run install.sh, or install.ps1 on Windows. It covers management commands, bounded loops, the local dashboard and Brain integration. Apache-2.0 licensed.