Your coding agents, always in the conversation.
NightMe connects your local coding agents to the chat apps you already use. Run Claude Code, Codex, OpenCode, Pi and more with one consistent workflow—while keeping your projects, sessions, and agent choices under your control.
Your agents can change. Your workflow doesn't have to.
Claude Code today. Codex tomorrow. A different agent for the next task. NightMe keeps the way you work consistent while letting you choose the agent that fits the job.
Agent-agnostic by design
NightMe does not lock your workflow to one vendor, one model, or one remote-control app. It connects the local agents you already use to the same conversation and control layer.
Not another AI model. Not another coding environment. NightMe is the control layer between your conversations and the coding agents running on your machine.
Code in the conversations you already use.
No new mobile IDE. No proprietary chat client to learn. Use a supported IM app you already know to talk to your local coding agents through NightMe.
Send instructions and review responses using the chat experience you already understand.
NightMe is designed to work across supported chat platforms instead of forcing you into a single messaging product.
Statusbar and reaction signals keep you updated while the local agent keeps working.
Inside one conversation
Every message routes to the local agent in the right project, and replies come back with full context.
[Lz] NightMe
- • [ ] Research release assets and current install UX
- • [ ] Design installer architecture and edge cases
- • [ ] Write html/install.sh (POSIX bash)
- • [ ] Write html/install.ps1 (Windows PowerShell)
- • [ ] Sanity-check and report results
Your coding sessions stay alive.
Close the terminal. Leave the desk. Come back later. Your NightMe session stays tied to its project so you can continue the same work instead of starting over from a blank context.
The session runs inside its CWD, and new messages continue to target that project until you decide to switch.
NightMe is built around persistent agent sessions, not one-off commands that vanish when the connection closes.
Start, stop, steer, and continue the work yourself.
Choose the agent + harness that fits the work.
Different tasks need different execution environments. Use the agent and harness that best fit the work—Claude Code for coding, Codex for writing, Pi for other workflows—while keeping the same NightMe conversation and control experience.
claude-code
harness: cli
codex
harness: cli
pi
harness: cli
Choose the Agent + Harness combination that gives you the most effective environment for the task instead of forcing every task through one tool.
Different conversations can represent different projects, tasks, and agent sessions at the same time—each with its own working directory.
Different tasks. Different agents. One consistent workflow.
Let it run. Know when it needs you.
NightMe keeps you informed while your agent works. Status updates and reaction signals tell you what is happening, so you don't have to keep checking in or babysit the agent.
See the session's current state without asking for an update.
Get clear feedback from the agent as it works, so you know when it is running, waiting, or finished.
/stop ·
/steer
— change direction when you need to, then let the agent continue.
The agent keeps working. NightMe keeps you informed. You step in when it matters.
From conversation to pull request.
When the work needs to become a clean Git change, /gtw connects coding agents with worktrees and pull requests—so parallel fixes can stay isolated from the main working tree.
-
1
Isolated worktrees
Give each task its own branch and working directory.
-
2
Agent-powered fixes
Start focused coding work without disturbing the main project session.
-
3
Ready to ship
Finish the task, review the result, and create or close the workflow cleanly.
Keep the conversation simple. Keep the Git state safe.
# Auto-rehydrate dev environment # in the newly created worktree fix: hooks: after: - codegraph init - npm install - go mod download
Bring the coding agents you already use.
NightMe does not replace your agent or your model subscription. It gives different local coding agents the same conversation and control layer, so you can switch tools without switching your workflow.
Use the best agent for the task. Keep the same NightMe workflow.
From zero to your first remote coding session.
-
Step 1
Install NightMe
Install the single binary on the machine where your coding agents run.
-
Step 2
Connect your chat
Sign in to a supported chat platform and connect it to NightMe.
-
Step 3
Choose your agent
Start the local coding agent that fits the task. Your NightMe workflow stays the same.
-
Step 4
Send your first task
Ask the agent to work. Leave the desk. Follow the progress from the conversation.
-
Step 5
Step in when it matters
Stop or steer the session when you need to change direction.
# One line. Installs everything. You're welcome. 🌙
$ curl -fsSL https://nightme.dev/install.sh | bash
curl -fsSL https://nightme.dev/install.sh | bash
# One line. Installs everything. You're welcome. 🌙
$ powershell -c "irm https://nightme.dev/install.ps1 | iex"
powershell -c "irm https://nightme.dev/install.ps1 | iex"
# Pick the asset for your OS + arch from the releases page.
# Pick the .exe for your arch from the releases page.
# Build from source. make dev runs from source with example config — no separate build step.
$ git clone https://github.com/cnlangzi/nightme.git
$ cd nightme && make dev
git clone https://github.com/cnlangzi/nightme.git
cd nightme && make dev
# Build from source. Run inside Git Bash or WSL.
$ git clone https://github.com/cnlangzi/nightme.git
$ cd nightme; make dev
git clone https://github.com/cnlangzi/nightme.git
cd nightme; make dev
Your agents can change. Your workflow can stay the same.
Run the coding agent that fits the job, keep every project and session close at hand, and stay in the conversation wherever you are.