AI coding session managers for applied AI researchers, compared

Q: Is AgentHydra free to use?

AgentHydra is free, with no account or subscription. It is open source under the MIT License. Its cost view shows figures at API list prices by day, model, project and account. Check how each figure is calculated before you rely on it for a budget decision.

Compare tools that list, track and report on AI coding sessions, with questions applied AI researchers should ask before choosing one.

By SupaKeeper · 2026-11-02

SupaKeeper is published by LunarWerx, which makes some of the tools compared here.

Pick by the job first, since these tools cover different parts of the work. AgentHydra shows Claude Code, Codex and OpenCode sessions on your machine in one searchable list. ^[1] It is free, with no account or subscription. ^[1] Each option's details below come from its own published source.

Option	What it does	Install or platform as stated
AgentHydra	Shows Claude Code, Codex and OpenCode sessions on your machine in one searchable list, updated live while they run. ^[1]	On Windows, install with a one-line PowerShell script, or download a build from GitHub Releases or run from source with Bun. ^[1] The tray app needs Windows; macOS and Linux builds exist but have not been verified. ^[1]
Claude Squad	A terminal app that manages multiple Claude Code, Codex and Gemini sessions, plus other local agents including Aider, in separate workspaces. ^[2]	Can be installed via Homebrew. ^[2]
OpenCode	An open source AI coding agent. ^[3] It supports multiple sessions with agents running in parallel on the same project. ^[4]	Available as a terminal-based interface, desktop app, or IDE extension. ^[3]
opcode (formerly Claudia)	Described on its page as a GUI app and toolkit for Claude Code. ^[5]	Requires Windows 10/11, macOS 11+, or Linux (Ubuntu 20.04+). ^[5]
ccusage	Analyzes coding agent CLI token usage and costs from local data. ^[6]	Can be run directly without a global installation, via npm or other package runners. ^[6]

What applied AI researchers need from AI coding session managers

Applied AI researchers often run several coding sessions at the same time, with one agent writing code, another reading results and a third checking a change. Finding the right session quickly matters when the work is spread that way, because a session you cannot locate is a session you cannot reuse. Check which tool shows the quota or budget each session uses. Cost by day, model or project can help decide what to run next. Reproducibility is another need: returning to an earlier session, seeing what changed and keeping the commands that worked. Ask whether your sessions run on one machine or across several, since that changes which tools are worth comparing. Decide which of these jobs matters most, then check whether each option covers it.

AgentHydra for applied AI researchers

AgentHydra is made by the team that publishes this site. It shows Claude Code, Codex and OpenCode sessions on your machine in one searchable list, updated live while they run. ^[1] You can type a reply into a Claude session's own desktop chat, or send one message to several sessions. ^[1] It can split a task's pieces across Claude Code CLI sessions on your accounts, and move a worker to another account when one nears its limit. ^[1] For cost, it shows figures at API list prices by day, model, project and account. ^[1] The same cost numbers can be printed as JSON from the command line. ^[1] It is free, with no account or subscription. ^[1] It is open source under the MIT License. ^[1] On Windows, install with a one-line PowerShell script, or download a build from GitHub Releases or run from source with Bun. ^[1] The tray app needs Windows; macOS and Linux builds exist but have not been verified. ^[1]

Claude Squad for applied AI researchers

Claude Squad is a terminal app that manages multiple Claude Code, Codex and Gemini sessions, plus other local agents including Aider, in separate workspaces. ^[2] It uses tmux to create isolated terminal sessions for each agent. ^[2] It can be installed via Homebrew. ^[2] Check whether the agents you run are among the ones named, and whether separate terminal sessions suit how you follow several experiments at once.

OpenCode for applied AI researchers

OpenCode is an open source AI coding agent. ^[3] It supports multiple sessions with agents running in parallel on the same project. ^[4] It is available as a terminal-based interface, desktop app, or IDE extension. ^[3] If you run several experiments on one project, check how the interface you pick shows each session, and whether that view suits how you compare results.

opcode (formerly Claudia) for applied AI researchers

opcode is described on its page as a GUI app and toolkit for Claude Code. ^[5] Its page lists requirements of Windows 10/11, macOS 11+, or Linux (Ubuntu 20.04+). ^[5] It also says its release executables will be published soon, so check the release status before you plan an install. ^[5]

ccusage for applied AI researchers

ccusage analyzes coding agent CLI token usage and costs from local data. ^[6] It can be run directly without a global installation, via npm or other package runners. ^[6] Check whether usage and cost reports are what you need, or whether you also want to manage sessions.

Matching the tool to the job

To find every running session in one searchable place, AgentHydra covers Claude Code, Codex and OpenCode sessions on your machine. ^[1] To reply to a Claude session from one place, AgentHydra lets you type a reply into a Claude session's own desktop chat. ^[1] To run several agents on one project at the same time, OpenCode supports multiple sessions with agents running in parallel on the same project. ^[4] To work across several agents in separate workspaces, Claude Squad manages multiple AI agents in separate workspaces. ^[2] To see what a set of runs cost by day, model or project, AgentHydra shows cost at API list prices by day, model, project and account. ^[1] ccusage analyzes coding agent CLI token usage and costs from local data. ^[6] To keep a task moving when one account is close to its limit, AgentHydra can move a worker to another account. ^[1] To use a graphical app around Claude Code, opcode is described on its page as a GUI app and toolkit for Claude Code. ^[5]

Questions to ask before you choose

Do your sessions run on one machine or across several accounts, and does the option you pick see both? Do you need cost by day, model or project, or is a session list enough? Which operating system will you use, and does the install route for it appear on the option's page? Do you want to reply to several sessions from one place, or keep each session separate? Do you run agents from more than one vendor, and does the tool name each one you use? Do you already use a tool that covers part of this job, and would a second one overlap with it?

Frequently asked questions

Does AgentHydra show Codex sessions in one list?

Yes, AgentHydra shows Claude Code, Codex and OpenCode sessions on your machine in one searchable list, updated live while they run. ^[1] It is free, with no account or subscription. ^[1] If you plan to use it on macOS or Linux, check the platform notes first, since the tray app needs Windows. ^[1]

Can OpenCode run several sessions on the same project?

OpenCode supports multiple sessions with agents running in parallel on the same project. ^[4] It is an open source AI coding agent. ^[3] It is available as a terminal-based interface, desktop app, or IDE extension. ^[3] Check that the interface you choose matches how you start sessions and follow their results.

Is AgentHydra free to use?

AgentHydra is free, with no account or subscription. ^[1] It is open source under the MIT License. ^[1] Its cost view shows figures at API list prices by day, model, project and account. ^[1] Check how each figure is calculated before you rely on it for a budget decision.

How does ccusage report coding agent costs?

ccusage analyzes coding agent CLI token usage and costs from local data. ^[6] It can be run directly without a global installation, via npm or other package runners. ^[6] Because it works from local data, confirm that the agents you use keep the records it needs on your machine before you depend on its reports.

Which operating systems does opcode support?

opcode requires Windows 10/11, macOS 11+, or Linux (Ubuntu 20.04+). ^[5] Its page says release executables will be published soon, so check that a build is available for your system before you plan an install. ^[5] It is described on its page as a GUI app and toolkit for Claude Code. ^[5]

Free, with no account or subscription. ^[1]

Try AgentHydra