Every agent, every subscription, every computer. One window.

Who is waiting on you, what each subscription has left, what each pull request cost.

No account, no telemetry. Between your computers, end-to-end encrypted.

Reads Claude Code, OpenCode, Codex, Gemini CLI, DeepSeek Harness and Cursor.

Caprock is a desktop app: it installs on your computer, not your phone.

Or email yourself the link:

macOS, Windows and Linux. Your phone pairs with it by QR code. Install options →

Download on the App StoreDownload on the App Store

On an iPhone, pair it with the app.

Download for Windows

Windows 10 22H2 or later, x64. Windows may ask once whether to run it: click More info, then Run anyway.

Paste it into Claude Code, Codex or any coding agent.

See what it says ▾
Install the Caprock desktop app for me. It is open source (Apache-2.0), a local
app for my coding agents: https://github.com/dspv/caprock

What the install does and how to remove it:
https://github.com/dspv/caprock/blob/master/docs/install-agent.md

Run the install script for my system: https://caprock.dev/mac.sh, /win.ps1 or
/linux.sh. Read it before you run it if you like. Leave my Claude Code settings
alone; the app asks me itself. Tell me when it's open.
All install options →
caprock· all computers

Mac mini 1 needs you

  • Make the webhook sender retry Claude wants to run go test ./internal/sender/... needs you
  • Backfill the orders table Claude Code working

This MacBook

  • Split the invoice PDF renderer Claude Code working
  • Port settings to the new form Codex working

Office PC

  • Run the end-to-end suite Codex working

Subscriptionson this MacBook

Claude · Work default

5-hour 38% left

Claude · Personal

5-hour 81% left

Codex

week 92% left

Three computers and three subscriptions in one window. Example figures.

Or run it yourself

curl -fsSL https://caprock.dev/mac.sh | sh

Paste in Terminal: it installs Caprock and opens it.Or with Homebrew: brew install --cask dspv/tap/caprock-app

irm https://caprock.dev/win.ps1 | iex

Paste in PowerShell: it installs Caprock for you and opens it.

curl -fsSL https://caprock.dev/linux.sh | sh

Paste in a terminal: it installs Caprock for you, no sudo. Or download the AppImage

FAQ

Questions.

What does Caprock do?

Caprock is a free, open-source desktop app for macOS, Windows and Linux that keeps the ledger of every coding agent you run, in one window — Claude Code, Codex, OpenCode, Cursor and others, the sessions you start in your own terminal included, and the Gemini CLI sessions it starts, on this computer and on the other computers you link: who is waiting on you, who is stuck, how much of its usage limit each Claude and Codex subscription has left, and what each pull request cost. Caprock records every session on the machine whether or not it started it, across Claude Code, Codex, OpenCode, Hermes Agent and DeepSeek Harness (Gemini CLI when Caprock starts it, and Cursor's chats with the cost Cursor recorded), prices each turn at list price when it is written, and joins it to the repository, pull request, CI run and release it produced. Run your agents in Caprock's terminal tabs and split panes, or keep your own terminal and your own phone: the ledger is the same either way. Other computers that run Caprock link to it and show their sessions in the same sidebar; several Claude Code and Codex accounts sit side by side, each with its usage limits; a paired phone answers permission prompts.

Is there a free app that tracks Claude Code, Codex and OpenCode in one place?

Yes. Caprock is a free, open-source (Apache-2.0) desktop app for macOS, Windows and Linux that records every Claude Code, Codex and OpenCode session on your machine in one dashboard, with Cursor's chats and the Gemini CLI sessions it starts: usage and spend per repository (at API list prices, and Cursor's at the cost Cursor recorded), Claude Code's and Codex's plan limits, and a history you can search. It reads Claude Code's, Codex's and OpenCode's own transcripts or database, and Cursor's Agent and Composer chats with the cost Cursor recorded, sessions from before you installed it included, with no account and no telemetry. Usage trackers compared →

Is there a desktop app for Claude Code and Codex?

Yes, one from each maker: Anthropic's Claude desktop app runs Claude Code in its Code tab, and OpenAI's ChatGPT desktop app runs Codex. To run both side by side in one window, Caprock does it free on macOS, Windows and Linux, as do Orca, T3 Code, Nimbalyst and Warp, and Conductor on macOS (as of November 2026). Coding agent desktop apps compared →

Is there a GUI for Claude Code?

Yes. Anthropic's official GUI is the Code tab of the Claude desktop app (macOS and Windows, Linux in beta, on a paid plan), with a VS Code extension, a JetBrains plugin and claude.ai/code beside it. Caprock is a free, open-source GUI around the claude you already installed: each session in a terminal tab, beside Codex and OpenCode, every one recorded. Claude Code GUIs →

How can I run several coding agents in parallel on my machine?

Give each agent its own git worktree so their edits never collide: Claude Code does it with claude --worktree <name>, and Codex gives each desktop-app chat a worktree. Caprock runs Claude Code, Codex, OpenCode and Gemini CLI side by side in tabs and split panes, each in a worktree of its own if you choose, with the sessions waiting on you first. Run agents in parallel →

What is the best tool to monitor Claude Code usage and cost?

Inside Claude Code, /usage shows the current session's cost and tokens. Outside it, it depends on the question: ccusage for reports on demand, Claude-Code-Usage-Monitor for the five-hour window live, CodexBar for limits in the Mac menu bar, tokscale for many agents in one terminal view, and Caprock for every session on the machine, whichever terminal started it, with plan limits and cost per repository at API list prices. Claude Code usage monitors →

How can I see how much I spend on Claude Code per project?

Claude Code has no per-project cost view: /usage covers the current session. ccusage reads the logs Claude Code keeps per project folder (ccusage claude daily --instances), and Caprock shows cost per repository, and per service in a monorepo, for Claude Code, Codex and OpenCode. On a flat plan these are what the same tokens would cost at API list prices, not a bill. Cost per repository →

How do I see my Claude Code and Codex usage limits in one place?

Caprock shows both side by side: Claude's 5-hour and weekly windows with when each resets and when the current pace reaches 100%, and Codex's windows as Codex last wrote them into a session. With several accounts, each is listed with what is left. On their own, Claude Code's /usage has the plan's usage bars on Pro, Max and Ultra, and /status in a Codex CLI session shows what is left of Codex's limits. How to check Claude Code usage →

How do I see what a pull request cost in agent tokens?

Caprock prices each pull request once GitHub is connected: the agent sessions on its branch at API list price (a session that worked on several pull requests is split between them), with the tokens behind the figure, an estimate rather than a bill. The Pipeline shows it from open through CI, merge and release, and sums it for each release. Pull requests and releases →

How do I search my old Claude Code and Codex sessions?

Each agent's resume picker finds a session in the list: claude --resume, and codex resume --all for every folder. To search what was said, run rg over ~/.claude/projects and ~/.codex/sessions; Claude Code deletes its own transcripts after 30 days by default. Caprock keeps every Claude Code, Codex and OpenCode session in a local database and searches what the agents wrote, after the files are gone. Claude Code history →

Can I control Claude Code from my phone?

Yes. Anthropic's official way is Remote Control: run /remote-control in a session and continue it in the Claude app for iOS and Android or at claude.ai/code, on a paid plan and not with an API key. Caprock pairs a phone by one QR scan, through Caprock for iPhone or any phone's browser, on any plan, and lets it answer permission prompts, type into and stop the sessions Caprock started. Claude Code on your phone →

Can I use multiple Claude or Codex accounts in one app?

Yes. Caprock keeps several Claude Code and Codex accounts signed in side by side, each with its own usage limits: Claude's 5-hour and weekly windows, and the windows Codex reports. You choose which subscription a new session uses. When a Claude Code session Caprock started hits its limit, you can continue it on another Claude account as the same conversation. Three Claude Code and three Codex accounts are free, counting the one you already use; Premium adds more. Each account is the agent's own login, kept in a separate folder, and Caprock never sees a password or a token. Claude Code usage limits →

How is Caprock different from Claude Code Remote Control?

Caprock is a separate, free app around the Claude Code and Codex you already pay for. It lists the sessions on each computer you link in one sidebar and shows the usage limits of the accounts on the computer you are using. It reaches your other computers and your phone through an end-to-end encrypted relay, with no Caprock account. Remote Control is Anthropic's own feature for Claude Code: the session keeps running on your computer and you continue it at claude.ai/code or in the Claude app for iOS and Android. It is available on Pro, Max, Ultra, Team and Enterprise plans, not with an API key, and while it is connected the transcript is stored on Anthropic's servers (Anthropic's documentation, read 20 November 2026). Either way, Claude Code itself talks to Anthropic as it does in any terminal. Claude Code Remote Control →

Can I run Claude Code on a Mac mini, a home server or a VPS and control it from my laptop?

Yes, with Caprock on both computers and the agent installed and signed in on the Mac mini, home server or VPS. From the laptop you see its Claude Code sessions and start an agent there; for the sessions Caprock started there, you also type into them and answer their permission prompts. Caprock's daemon runs without the app on macOS, Linux and Windows, set up over SSH on a computer with no screen. The link is end-to-end encrypted, with no Caprock account, and linking computers is free, with no limit on how many. Install without the app →

How many Claude Code sessions can I run at once?

Caprock sets no limit on how many Claude Code, Codex or other agent sessions run at once; the computer's memory does. The status bar shows this computer's memory and each agent's share, and one button ends idle agents and keeps their conversations.

Is there a Claude Code desktop app for Linux?

Yes, in beta: Anthropic's Claude desktop app, with Claude Code in its Code tab, runs on Ubuntu 22.04 or later and Debian 12 or later, on a paid plan. Caprock's free app runs the claude you installed in terminal tabs on current x64 distributions with WebKitGTK 4.1, as an AppImage, .deb or .rpm. Claude Code on Linux →

What are alternatives to Conductor for running Claude Code agents?

The closest are Orca and T3 Code, free and open source on macOS, Windows and Linux, and Paradev on a Mac; Nimbalyst and Emdash also give each agent its own git worktree. Caprock, free and Apache-2.0, runs Claude Code beside Codex and other agents and records every session with its cost. Conductor itself is not shutting down, and runs on macOS only (as of 18 November 2026). Conductor alternatives →

How do I check my Codex usage limits?

OpenAI's usage dashboard at chatgpt.com/codex/settings/usage shows your current Codex limits and reset times, and OpenAI's pricing page points to /status in a Codex CLI session for what is left. Caprock shows Codex's windows beside Claude's, for each Codex account you sign in with. Codex usage limits →

Is there a desktop app and a phone app?

Both. On an iPhone, Caprock is a free app on the App Store (iOS 15 or later) that pairs with the desktop app by scanning its code; on any other phone it is a page in the browser. The desktop app runs on macOS (download it, or brew install --cask dspv/tap/caprock-app; signed with an Apple Developer ID and notarized by Apple), Windows and Linux, with the daemon — the background service that records your sessions — inside. It opens on your agent terminals and projects, and when a Claude Code session Caprock started asks for permission, the notification on macOS and Linux has Deny, and Approve when it shows the whole request. A phone pairs by QR code and opens Caprock in its browser, from anywhere, through an end-to-end encrypted relay: it reads a session as a chat and reconnects by itself, and once you allow control it answers prompts, clones a project and starts an agent in it.

Do I have to change how I work?

No. Keep running claude in your own terminal. Caprock adds a small hook to ~/.claude/settings.json — backed up first, removed with caprock hooks uninstall — and reads the transcripts Claude Code already writes.

Will it see sessions from before I installed it?

Yes. Caprock imports transcripts still on disk, which Claude Code keeps for 30 days by default. It then records new sessions and keeps them after Claude Code deletes the original files.

Does anything leave my machine?

Only if you enable a network feature. By default, data stays in a SQLite file on your disk and Caprock listens on 127.0.0.1. There is no account or telemetry. You can enable update checks (the app asks once), connect GitHub to clone and open pull requests, send weekly reports and phone alerts to your Telegram bot, or ask Gemini using your API key. Updating downloads the new app from GitHub's release files. You can also pair a phone, or link another of your computers: either reaches Caprock from anywhere, through an end-to-end encrypted relay that sees which machines talk, when and how much, never what. A phone you pair can control sessions unless you untick that when you pair it, and you can take control away in settings. A bug report or a survey answer reaches us only when you press Send. Pressing X, Threads or LinkedIn on the Week card uploads the picture you chose, alone, to caprock.dev so the post shows it; Settings → Privacy lists each one with Remove. The code is Apache-2.0 on GitHub.

How does it spot a stuck session?

When a session calls the same tool with nearly the same input five times within three minutes, a banner goes up. Repeated reads and searches don't count: re-reading a file is normal work, and on 64,733 real tool calls they were over half of all alerts. Sessions Caprock started can be paused automatically.

Is it free?

Yes, for solo use, and what is free stays free: live sessions, cost per repository, plan-limit warnings, history and search. A licence — $30 a year or $100 once — adds 8 features: Auto-pause at a daily budget, Pause at the plan limit, Answer from Telegram, A weekly report, Ask about your usage, Where each tool failed, Custom agents, More than three sign-ins per agent. The desktop app is free too. Caprock adds nothing to your Claude bill. The costs it shows are at API list prices: on a flat plan, an equivalent, not a charge.

What are good alternatives to ccusage?

Keep ccusage for a report when you ask for one: it prints usage tables in your terminal, with nothing running in between. For something that stays open: tokscale for a terminal view, CodeBurn for a desktop and menu-bar app, CodexBar for plan limits in the Mac menu bar, or Caprock, which runs in the background and keeps a ledger: live sessions, loop alerts, plan limits, cost per repository and per pull request, and history kept past Claude Code's 30-day cleanup. ccusage alternatives →

What is the task runner?

An opt-in way to run tasks unattended. You write a task and the commands that must pass, such as tests, type checks and lint. Each task runs in its own git worktree; Caprock runs your commands itself before marking the task done. It never merges into your branches. Workers skip permission prompts and your commands run with your permissions, unsandboxed — give it independent tasks you would run yourself.

The problem

What people who run many agents say.

  • Which one is waiting

    “No overview of all running sessions with notifications; I jump between tmux sessions/windows.”
    mainmind83 GitHub, claude-code #107064 3 Nov 2026
  • The laptop has to stay open

    “A laptop open on the desk overnight is the price of admission.”
    Jean Galea jeangalea.com 15 Jun 2026
  • A computer at home

    “I have an M4 Mac Mini running 24/7 so I can use it as my Claude Code machine when I'm on the go.”
    aurareturn Hacker News 1 Sep 2026
  • A second subscription

    “I could not wait two days. So I bought a second Max subscription.”
    intermap74 GitHub, claude-code #93799 12 Sep 2026

Quotes read 20 November 2026. They are about the problem, not about Caprock.

Computers

Run Claude Code and Codex on your other computers, from this one.

Caprock is the remote control for the agents on your other computers: see their sessions, start an agent there and answer its prompts, in one window.

Each tagged with its computer. The one waiting on you comes first.

Type into one and answer its permission prompts, for sessions Caprock started there.

Example figures.

The other computer can be a Mac, a Linux box or a Windows PC running Caprock. Set it up on a computer with no screen →

Agents on your other computers → · Claude Code Remote Control → · Codex remote control →

Subscriptions

Multiple Claude and Codex accounts, each with its usage limits.

Caprock shows what each account has left, side by side: Claude's 5-hour and weekly limits, and the limits Codex reports.

What is left of each, and when it resets.

Example figures.

Free: three Claude Code and three Codex accounts, counting the one you already use. Premium ($30 a year or $100 once) adds more.

Each account is the agent's own login, kept in a separate folder. Caprock never sees a password or a token, and no request goes through it. Claude Code usage limits → · Several subscriptions in Caprock →

Phone

The agent waiting on you, answered from your phone.

In the desktop app, a session waiting on you comes first. If it runs on another computer, the desktop notification names that computer.

A session's chat on a phone while dictating: the message “Once CI is green, merge it and delete the branch” in the field, the microphone on, the language chip EN and SendA session's chat on a phone while dictating: the message “Once CI is green, merge it and delete the branch” in the field, the microphone on, the language chip EN and Send
Say it instead of typing
Caprock on a phone: a session waiting on you, with Claude asking to run git push and Yes and No buttons, above the sessions still workingCaprock on a phone: a session waiting on you, with Claude asking to run git push and Yes and No buttons, above the sessions still working
Approve on the go
Releases on a phone: the next release of caprock with a ten-block progress bar per pull request — merged, CI running, ready, CI redReleases on a phone: the next release of caprock with a ten-block progress bar per pull request — merged, CI running, ready, CI red
The next release in your pocket

A paired phone. Example figures.

Download on the App StoreDownload on the App Store

Caprock for iPhone: free, iOS 15 or later.

  • Pair by QR code

    In the free iPhone app or any phone's browser. Works from anywhere.
  • Answer, dictate, start, stop

    Answer prompts and start an agent from the phone. Type into and stop the sessions Caprock started.
  • One phone, several computers

    Pair your phone with each computer; it switches between them.

Claude Code on your phone, every way →

Memory

See how much memory each agent uses.

Caprock's status bar shows this computer's memory and each agent's share. One button ends idle agents and keeps their conversations.

What a session costs the machine →

Example figures.

What did that feature cost?

Cost per pull request, with the tokens behind it.

The pull requests open now or merged this week that cost the most, each priced from the agent sessions on its branch. Not one total for the month.

At API list price: an estimate of what those tokens cost, not a bill. Example figures from a demo machine.

A real week, read on the maker's machine 16 Oct 2026: 45 pull requests merged into dspv/caprock in 7 days, $343.78 of agent cost — about $7.64 each; the dearest, #720, $71.13. At API list price.

On the Home screen, with GitHub connected.

Cost and limits

What each repository costs, and how close the limit is.

Every turn priced at API list price when it is written, however you pay, with your Claude and Codex plan windows beside it.

Cost across your computers is on one screen. A computer that did not answer is marked offline, never counted as $0.

In August I ran $5,898 of Claude Code on a $200 Max plan — 29× at API list prices.

Cost

Claude Code cost per repository — and per service in a monorepo.

A turn counts toward the directory of the last file it touched, so the rows add up to the repository.

127.0.0.1:22776
Projects30d
example-app
41.2M
$3,426
by files touched · share of this repository’s tokens
/app
25.1M 61% · $2,090
/.ai
2.8M 6.8% · $233
/app/tests
1.1M 2.8% · $96
outside the repository
10.7M 26% · $891

A turn counts toward the directory of the most recent file it touched, and keeps counting there until work moves elsewhere. Each turn goes whole to one row, so the rows add up to the repository’s total exactly.

Example figures.

How close the limit is

Six days to 99%, then the reset.

One Claude plan's seven-day window on the maker's Mac, read every hour by the app. The 5-hour window sits beside it in the app, with when each resets and when this pace reaches 100%.

Claude Code usage limits, plan by plan →

One Claude plan’s seven-day limit, 13 to 21 October 2026Per cent of the weekly limit used, read every hour. It climbs from 5% on 13 October to 99% on the morning of 19 October, drops to 1% at the reset at 16:00 that day, and reaches 29% by the end of 21 October.100%50099%reset131415161718192021Oct
One developer’s Mac, 13 to 21 October 2026: one Claude plan’s seven-day window, read every hour.

Pipeline

Every pull request, from agent to release.

Read from your git and GitHub, on your machine: where each pull request is, what the agents on its branch cost, and what went out in each version.

Open, CI, merged, released, and what the agents spent on it.

Example figures from a demo machine. Agent cost is at API list price, an estimate.

Work

Who is on what, how far it got, and who is stuck.

Every branch at its furthest stage, with the agent and model on it, how long it has run and what it has cost so far. A session calling the same tool over and over gets a loop alert.

Example figures from a demo machine. Agent cost is at API list price, an estimate.

Your agents

Run them here, or in your own terminal.

Claude Code, Codex, OpenCode, Gemini CLI and Hermes Agent in tabs and split panes, each in its project or worktree. Or keep your terminal: Caprock records those sessions too, and the ledger is the same.

⌘K to anything

Sessions waiting on you come first; a new agent is one line away.

Claude Code desktop app options →

Example figures.

What your agents shipped today.

Free and open source · macOS, Windows and Linux · no account

View on GitHub →

Caprock is a desktop app: it installs on your computer, not your phone.

Or email yourself the link:

macOS, Windows and Linux. Your phone pairs with it by QR code. Install options →

Download on the App StoreDownload on the App Store

On an iPhone, pair it with the app.

Download for Mac

Notarized by Apple. One file for Apple silicon and Intel, macOS 13 or later.

Download for Windows

Windows 10 22H2 or later, x64. Windows may ask once whether to run it: click More info, then Run anyway.

Paste it into Claude Code, Codex or any coding agent.

See what it says ▾
Install the Caprock desktop app for me. It is open source (Apache-2.0), a local
app for my coding agents: https://github.com/dspv/caprock

What the install does and how to remove it:
https://github.com/dspv/caprock/blob/master/docs/install-agent.md

Run the install script for my system: https://caprock.dev/mac.sh, /win.ps1 or
/linux.sh. Read it before you run it if you like. Leave my Claude Code settings
alone; the app asks me itself. Tell me when it's open.

Or run it yourself:

curl -fsSL https://caprock.dev/mac.sh | sh

Paste in Terminal: it installs Caprock and opens it.Or with Homebrew: brew install --cask dspv/tap/caprock-app

Or run it yourself:

irm https://caprock.dev/win.ps1 | iex

Paste in PowerShell: it installs Caprock for you and opens it.

Or run it yourself:

curl -fsSL https://caprock.dev/linux.sh | sh

Paste in a terminal: it installs Caprock for you, no sudo. Or download the AppImage

Private by design

Your sessions stay on your computers.

No account, no telemetry. Network features, a paired phone and a linked computer among them, are each off until you turn them on, and the app says what each one sends.

Your computer

127.0.0.1:22776caprock.db · SQLite

Off until you turn it on

Release check · offGitHub · offPhone · offComputers · offTelegram · off

Local by default

Data stays in a SQLite file on your disk. There is no account or telemetry. It goes online only for what you switch on: update checks, GitHub, a paired phone, a linked computer, your own Telegram bot, a week card you post. Each is listed where it is switched on, with what it sends.

~30 MB

The daemon, idle.

~40 MB while reading 1,414 transcripts. Measured 2026-08-18.

hook fails → exits in < 1 s

Hooks exit safely on error

If the hook fails, your session goes on.

your terminal · observedstarted by Caprock · controlled

You control your terminal sessions

Observes what you start; controls only what it starts.

priced from a dated list

No invented numbers

Every dollar is a list-price estimate, never a bill.

macOSWindowsLinux

Three operating systems

macOS, Windows, Linux. Tested on all three in CI.

Compared

Caprock and other apps for running coding agents.

Each runs agents in parallel. What differs is the price and licence, the platforms, how a phone reaches your machine, and what you see of cost and limits.

Feature Caprock Orca Conductor Claude app ChatGPT app Warp Nimbalyst
macOS, Windows and Linux ✓ ✓ — ◐ ◐ ✓ ✓
Phone with no account ✓ ◐ — — — — —
Cost per repository ✓ — — — — — —
Claude and Codex plan limits ✓ ✓ ✓ ◐ ◐ — ✓
Approve from a notification ◐ — — — — — —
System prompt per project ✓ ✓ ✓ ◐ ◐ — —
Free and open source ✓ ✓ ◐ — ◐ ✓ ✓

✓ yes · ◐ partly · — no, or not in their docs · read 21 October 2026 · All columns, with sources → · Claude Code GUIs → · Open-source agents →

Claude Code compared with Codex · Cursor · OpenCode · Gemini CLI · Copilot · Antigravity · Kiro

  • Caprock fits

    Several agents, more than one subscription, more than one computer: who is waiting, what each subscription has left, what each PR cost. In one window.
  • Look elsewhere

    You want an editor or cloud workspaces. Cloud and CI sessions are not observed.

Try it

On the sessions you already run.

Install the app and it picks up the sessions already on this machine, past ones too.

Caprock runs on macOS, Windows and Linux.Notarized by Apple, for Apple silicon and Intel.Windows may ask once: click More info, then Run anyway.Paste it as a new message in your coding agent. Other ways →

Caprock with six Claude Code agents in tabs, a shell under the running one, and its cost on the rightCaprock with six Claude Code agents in tabs, a shell under the running one, and its cost on the right

The ledger

Keep your coding agents’ history.

Caprock imports the sessions still on disk, not ones already deleted, and records new ones. You can search them after your agents delete the original transcripts.

127.0.0.1:22776
Last 30 days
$342.09
at API list prices · 3 models
opus-4.8 · $270
sonnet-4.6 · $50
haiku-4.5 · $22

Example figures.

  • Kept past 30 days

    Claude Code deletes transcripts after 30 days by default. On the maker's machine, 67 of 114 transcripts were gone, including 10 from worktrees he deleted. All 67 remained searchable in Caprock. The numbers →
  • Six agents, one table

    Claude Code, Codex, OpenCode, Gemini CLI, DeepSeek Harness and Hermes Agent each write their own format. Caprock reads all six into one table on your disk; caprock export gives it to you as TSV, CSV or JSON lines. How →
“not just tracking spend, but seeing what's really happening on your projects. … PyCharm restarts and the session is gone. Caprock takes that pain away.”
Vladimir Konov
Vladimir KonovSoftware Engineer · Rust, distributed systems read the whole note
“I see its future value as a central, independent store of my sessions across different models, providers and projects … use as parallel documentation or extra memory.”
Sergey Aksёnov
“Something like the Meyer Lansky of the agent world: it tirelessly makes sure every dollar goes where it's supposed to, and it's always ready to account for every cent … For startups counting every cent, I think it's a great solution.”
Aliaksandr Valynski
Aliaksandr ValynskiQA Engineer & System Analyst read the whole note

Every note in full, with how long each person used it →

For teams

What did your team’s agents ship this week, and what did it cost?

Every coding agent on every machine, on a server you own. Counted by pull request, not by person; prompts and code stay on each laptop.

caprockteam · merged PRs

Merged

58

GitHub

Per PR

≈$33

estimate

Unlinked

18%

$420

  • #482 Retry failed payouts$46.20payments-core · 3 sessionsby commit
  • #479 Apple Pay in checkout$38.10web-client · 2 sessionsby commit
  • #477 Incremental orders model$71.40data-pipeline · 4 sessionsby commit
  • 55 more merged PRs$1,734.30

Example figures, one team of eight.

From $500 a month, billed yearly — one flat price for the team, never a share of your AI spend. The Community Edition stays free.

Your agents are waiting

See every coding agent you run.

Who is waiting on you, what each subscription has left, what each pull request cost. Free and open source; no account, no telemetry.

Caprock is a desktop app: it installs on your computer, not your phone.

Or email yourself the link:

macOS, Windows and Linux. Your phone pairs with it by QR code. Install options →

Download on the App StoreDownload on the App Store

On an iPhone, pair it with the app.

Download for Mac

Notarized by Apple. One file for Apple silicon and Intel, macOS 13 or later.

Download for Windows

Windows 10 22H2 or later, x64. Windows may ask once whether to run it: click More info, then Run anyway.

Paste it into Claude Code, Codex or any coding agent.

See what it says ▾
Install the Caprock desktop app for me. It is open source (Apache-2.0), a local
app for my coding agents: https://github.com/dspv/caprock

What the install does and how to remove it:
https://github.com/dspv/caprock/blob/master/docs/install-agent.md

Run the install script for my system: https://caprock.dev/mac.sh, /win.ps1 or
/linux.sh. Read it before you run it if you like. Leave my Claude Code settings
alone; the app asks me itself. Tell me when it's open.

Or run it yourself:

curl -fsSL https://caprock.dev/mac.sh | sh

Paste in Terminal: it installs Caprock and opens it.Or with Homebrew: brew install --cask dspv/tap/caprock-app

Or run it yourself:

irm https://caprock.dev/win.ps1 | iex

Paste in PowerShell: it installs Caprock for you and opens it.

Or run it yourself:

curl -fsSL https://caprock.dev/linux.sh | sh

Paste in a terminal: it installs Caprock for you, no sudo. Or download the AppImage

Free and open source. Premium adds a daily budget stop and Telegram.

Teams: every machine, by repository and pull request, on your servers.