project
One durable scope
A repository or folder, its connections, tasks, sessions, and launch configuration—kept together without hiding the filesystem.
project → tasks + sessionstermloop · native control room for coding agents
Keep every coding agent in the loop
TermLoop puts tasks, worktrees, real terminals, diffs, and agent coordination in one native workspace—without replacing Claude Code or Codex.
available nowopen sourceyour subscriptionsno telemetry
tree project
Simple objects. Serious workflows.
Projects hold the durable scope, Tasks describe work, and Sessions are the live agents and terminals operating inside that Project.
task
Work with optional isolation
A Task can stay lightweight or get one managed Git worktree when the work needs its own branch and checkout.
task → 0..1 worktreesession
Every process stays visible
Claude, Codex, terminals, and dev runs stay Project-scoped. Select them, refresh them, fork context, or coordinate them directly.
session → agent | terminal | runfeature guide · 32 demos
Find your next better workflow.
Explore 32 explained desktop demos, including 19 new recordings. Launchpad is a working demo Project with source files, release tests, Tasks, and real Git changes. Open a guide for the steps, or expand a video to see the details.
32 features
Run
01 — Launch and isolate workquick actions
Start the right agent with the right context.
Open the command palette, compose a free prompt, choose the agent, model, permissions, and reasoning, then launch an inspected Session in seconds.
How to use Quick Actions
- Open Quick Action with Shift Shift.
- Describe the change and choose the agent settings.
- Launch a Session with the context attached.
task worktrees
Isolation when the task needs it.
Create a Task first, then provision one managed worktree when implementation should happen on its own branch. The Task stays useful even without a checkout.
How to use Task Worktrees
- Create a Task for one focused change.
- Give the Task its own branch and managed worktree.
- Keep parallel changes in separate checkouts.
dev server
Define it once. Start it where you work.
Save a Project dev-server command and fallback URL, then launch the real process from the Project or its Task worktrees. Its terminal and discovered URLs stay visible.
How to use Dev Server
- Save a preview command for the Project.
- Start the command in the checkout you are using.
- Follow the process output and discovered preview URL.
task briefs
Start with a clear definition of done.
Give each Task a title, a focused brief, and acceptance criteria. Open its detail view to keep the goal beside Sessions and checkout state.
How to use Task briefs
- Open the onboarding Task from the sidebar.
- Read its brief and concrete acceptance criteria.
- Keep the goal attached to the work as the Task evolves.
run configurations
Keep launch commands with the Project.
Save named commands for preview servers, tests, and other local processes. Configure setup commands and fallback URLs in the same editor.
How to use Run configurations
- Open the saved Launchpad preview configuration.
- Review pnpm dev, the setup command, and fallback URL.
- Reuse the configuration when starting work in a checkout.
Coordinate
02 — Agents working togethermulti-agent
One workspace. Several live agents.
Run multiple Claude and Codex Sessions across tasks and worktrees. See who is active, which checkout changed, and which terminal needs attention.
How to use Multi-Agent
- Keep several agent Sessions in one Project.
- Use the sidebar to see which Session needs attention.
- Open the right terminal without losing the other Sessions.
ask to
Get a second opinion without copy-paste.
Ask one agent to consult another provider. TermLoop creates a visible helper Session, preserves the relationship, and returns the answer to the requester.
How to use Ask To
- Ask the current agent for a second opinion.
- A visible helper Session receives the question and context.
- The answer returns to the requesting agent.
handoff
Move context to the agent already in the work.
Send a visible, composed handoff from one running Project Session to another. The target receives the brief in its own real terminal and continues from there.
How to use Handoff
- Choose the existing Session that should receive the brief.
- Compose the context and the next action.
- The destination Session receives the handoff in its terminal.
session fork
Explore another direction in a real terminal.
Fork a Session from the sidebar, see the parent-child relationship, and continue in a new Ghostty terminal. The branch in thinking is explicit instead of becoming another anonymous tab.
How to use Session Fork
- Choose a conversation to explore in a new direction.
- Fork it from the Session menu.
- Continue in a separate Session with visible parentage.
agent library
Choose a specialist for the next question.
Browse reusable agent profiles with a defined purpose. Inspect their provider, working mode, reasoning, and instructions before running one.
How to use Agent library
- Open the Agents library.
- Inspect the Edge Case Hunter profile and its working mode.
- Choose a profile whose instructions match the task.
custom agents
Save the reviewer your team actually needs.
Create a reusable agent with its own name, provider settings, and instructions. The Launchpad example checks progress bounds, tests, and keyboard accessibility.
How to use Custom agents
- Start from a duplicate of an existing specialist.
- Give the Launchpad reviewer focused, project-specific instructions.
- Save the profile so the same brief can be used again.
workflow templates
Make the collaboration repeatable.
Define discussion, implementation, review, and fix steps in a Project template. Choose the agent conversation and instructions for each step before running it in a ready Task worktree.
How to use Workflow templates
- Open the saved Discuss, build, review template.
- Inspect the ordered discussion, implementation, and review steps.
- Reuse the template for the next Task with a ready worktree.
review cycles
Give review a clear stopping rule.
Configure review and fix steps together, including a maximum number of review rounds. The template makes the review instructions and return path visible.
How to use Review cycles
- Select the accessibility and tests review step.
- Give the reviewer a specific checklist to assess.
- Bound the fix-and-review cycle with a maximum round count.
Review
03 — Understand and retain the workchanges
Review the diff where the work happened.
Changed-file counts appear on Tasks and active Agents. Open the Git changes view, inspect the real diff, and leave line-level notes for the agent to address.
How to use Changes
- Open the changed files for the checkout.
- Read the diff and attach a note to a relevant line.
- Send review notes to the selected active agent.
task lifecycle
Close the work without erasing it.
Tasks have a deliberately small lifecycle: open or closed. Close finished work, keep its current state visible, and reopen it when the next change arrives.
How to use Task Lifecycle
- Close a Task when the work is finished.
- Keep completed work in the Closed section.
- Reopen the Task when another change is needed.
session refresh
Refresh the display, not the identity.
Recover a stale agent display from the Session menu while keeping the same durable Session and process model. Lifecycle actions remain explicit.
How to use Session Refresh
- Open the Session actions menu.
- Refresh the provider display for the same conversation.
- Return to the durable Session after the display restarts.
developer notes
Keep small follow-ups beside the Task.
Turn the human checks around a change into a short checklist. Mark individual notes complete while keeping their context visible in the sidebar.
How to use Developer notes
- Start with the onboarding Task and its three checks.
- Mark the completed tests and keyboard check as done.
- See the checklist reach three of three without leaving the Task.
task favorites
Keep important work easy to recognize.
Mark a Task as a favorite directly from its row. The explicit favorite state lets you distinguish priority work from the rest of the Project.
How to use Task favorites
- Reveal the actions on the onboarding Task.
- Toggle the Task favorite with its star control.
- Keep the Task marked while its notes and brief stay intact.
full-file review
Read the context around the diff.
Switch from changed lines to the whole file when more context helps. Use unified or split comparison to inspect the same checkout changes.
How to use Full-file review
- Open the actual changes in src/releases.mjs.
- Switch from Change focus to Full file.
- Choose Split to compare the old and new versions side by side.
review progress
Remember which files you have checked.
Mark changed files as reviewed as you work through the diff. A visible counter shows what remains in the current review.
How to use Review progress
- Start with two changed files in the Launchpad checkout.
- Mark each file after inspecting its changes.
- See two of two reviewed before returning to the Task.
task archive
Park context without deleting the Task.
Use the archive preview to inspect whether a Task can be parked safely. Eligible Tasks move out of active work and remain available in Archived items.
How to use Task archive
- Choose Archive Task from the Task actions.
- Read the archive preview before confirming.
- The Task moves out of the active list and into the archive.
restore archived work
Bring a parked Task back into the Project.
Expand Archived items and restore the Task when work resumes. Its brief and developer notes return with it; eligible agent recovery follows the archive restore flow.
How to use Restore archived work
- Expand Archived items to find the onboarding Task.
- Restore the Task from its archived row.
- Its brief and completed developer notes return to active work.
Operate
04 — Keep the workspace runningsteward
A Project-scoped operator that stays present.
Ask the persistent Steward what should ship next, let it coordinate Tasks and Sessions through its constrained tools, and keep the conversation attached to the Project.
How to use Steward
- Open the Project Steward workspace.
- Keep project-level instructions beside the work.
- Coordinate Tasks through a persistent Project assistant.
mobile companion
Check the loop when you leave the desk.
Pair your phone, keep Project visibility and terminal input scoped to your computer, and follow long-running agents over your trusted network.
How to use Mobile Companion
- Open Connect Mobile from the desktop.
- Use the pairing flow on your trusted network.
- Continue with project visibility and terminal input on your phone.
mcp & prompts
Inspect the tools and instructions agents receive.
Browse role-scoped MCP tools and the built-in prompt catalog. TermLoop-generated instructions remain visible and editable; provider-managed opaque layers are labeled honestly.
How to use Mcp & Prompts
- Open the tools and prompt libraries.
- Read the instructions and tool descriptions agents receive.
- Keep TermLoop-authored context inspectable.
pause to read
Read output while the Session keeps running.
Pause the terminal display to inspect output at your own pace. Return to live when you are ready; pausing the view does not stop the process.
How to use Pause to read
- Run the five real Launchpad release tests.
- Use Pause to read to hold the displayed output.
- Return to live without ending the Session.
keep awake
Give long-running work time to finish.
Choose whether to keep the computer awake while agents run, continuously, or for a limited time. Keeping the display on is a separate option.
How to use Keep Awake
- Open Keep Awake from the sidebar.
- Compare Off, While agents run, and Always.
- Use a timer or the separate display option when needed.
notification preferences
Choose when each device should alert you.
Configure desktop, iPhone, and Apple Watch notification preferences. Separate input requests, review-ready alerts, and Steward messages by device.
How to use Notification preferences
- Open Settings and choose Notifications.
- Review alerts for agent input, completed turns, and Steward messages.
- Choose device-specific sound and active-computer behavior.
appearance
Match the workspace to your environment.
Choose Light, Dark, or System appearance. The resolved theme applies to the app and terminal panes, with System following the computer preference.
How to use Appearance
- Open Appearance in Settings.
- Compare the Light and Dark workspace themes.
- Choose System to follow the computer automatically.
Customize
05 — Give agents the right instructionscontext bank
See the instructions in your checkout.
Browse project and folder-level instruction files in a searchable tree. Open the real file, inspect its contents, and see its recommended line budget.
How to use Context Bank
- Open Context for the Launchpad Project.
- Read the root instructions, then the source-folder instructions.
- Use the file path and line budget to keep context focused.
instruction consistency
Resolve conflicting sibling instructions explicitly.
Context Bank flags differing sibling instruction files such as AGENTS.md and CLAUDE.md. Choose the source of truth and review the overwrite count before applying it.
How to use Instruction consistency
- Rescan after sibling instruction files change.
- Compare the source choices in the conflict resolver.
- Choose the source file and apply the explicit synchronization.
skill library
Find and inspect reusable skills.
Search discovered skills and open their actual instructions. The library shows their locations and provider deployment state so you can understand what is installed.
How to use Skill library
- Search the skill library for simplify.
- Open the skill and read its instructions in the editor.
- Inspect its discovered locations and provider installation state.
No matching features. Try a broader search or clear the filter.
product principles
The terminal stays real. The workflow gets better.
--no-replacement-agent
TermLoop launches the Claude and Codex CLIs you already trust and pay for. It never proxies or imitates them.
--no-hidden-filesystem
Branches, worktrees, processes, and files remain ordinary local resources you can inspect outside the app.
--no-mystery-prompts
TermLoop-authored prompts are visible assets. Secret values stay out of durable state and public evidence.
--no-fake-coordination
Agent-to-agent requests, handoffs, terminals, diffs, and lifecycle changes are real product actions.
straight answers
Frequently asked.
Which agents does TermLoop support?
The first-class agent launch paths are Claude Code and OpenAI Codex, alongside ordinary terminal and dev-run Sessions. TermLoop uses your existing provider sign-ins rather than proxying model traffic.
Does every Task create a worktree?
No. A Task can have zero or one worktree. Create the Task to track work; provision a managed checkout only when branch isolation is useful.
Can multiple agents write in the same worktree?
Yes. TermLoop does not impose a product-level write lease or artificial launch gate. You choose whether agents share a checkout or work in isolated Task worktrees.
Where do my code and credentials go?
Your code remains on your filesystem. Agents connect directly to their providers. TermLoop keeps secrets out of durable product state, logs, URLs, and generated evidence.
Which desktop platforms are available?
The current public desktop download is for macOS. TermLoop Next treats macOS, Linux, and Windows as release gates; broader availability follows verified native builds rather than an untested compatibility claim.
Is the mobile connection real?
Yes. The companion pairs with a running desktop instance for Project visibility and terminal input. The public demo intentionally masks the one-time QR payload.
Is TermLoop open source?
Yes. The public source and releases are available on GitHub, including the terminal and agent-workflow layers.
ready when you are
More agents. Less terminal babysitting.
Download TermLoop and keep the whole engineering loop visible.