harnsy

harnsy Max

Out of one subscription’s limits.

Your Claude, Codex, OpenCode and pi logins and API keys live in Connections, each project and each agent runs on the one you pick, and you switch when a limit runs out. It’s in every harnsy build; a harnsy Max key turns it on.

Connectionsharnsy Max
  • claude-personalWorkschecked 2 min ago
    ClaudeSubscription token…x9QaDefault for acme-shop
    5 hours92%resets in 1 h 28 min
    Week61%resets in 2 d 22 h
    $41.20 in 7 days2.4M in · 1M out12 sessions
  • codex-workWorkschecked 2 min ago
    CodexChatGPT login
    5 hours40%resets in 3 h 5 min
    Week22%resets in 4 d 5 h
    — in 7 days1.8M in · 420k out7 sessions
  • glmNot checked
    OpenCodeProvider keyZ.ai…m3Kc
    $6.80 in 7 days930k in · 210k out4 sessions
Connections in the dashboard: several logins on one machine, each with its limits and cost. Sample data.

Several logins on one machine.

Add Claude Code subscription tokens and Anthropic API keys, a ChatGPT login and OpenAI API keys for Codex, and provider keys for OpenCode, including other providers behind it, such as GLM on Z.ai. harnsy keeps the secrets encrypted and never shows them again.

Your own server for OpenCode.

Connect OpenCode to your own server with an OpenAI-compatible API, a local model for example. It is set up as a connection; a harnsy Max key turns it on.

A default for each project.

Each project has its own default connection, and you can start an agent on another one. When a limit runs out, move the project to another connection; an agent that is already running keeps its connection until it restarts.

Each connection at a glance.

A connection shows its 5-hour and weekly limits, its cost over 7 days, tokens and sessions. “Check” tests it; you can rename it, replace the secret, log in again or archive it.

The limits are in Usage.

Every limit with a forecast, its history and the spend live in Usage, the next page.

Usage →

Dictation is in both editions.

The speech-to-text providers for dictation sit on the same page and work without a key.

Dictation →

Project defaultsharnsy Max

An agent started in a project’s folder gets this connection unless another is picked at launch. Changing it does not move running agents: they keep theirs until restarted.

ProjectClaudeCodexOpenCode
acme-shopclaude-teamcodex-workHarness’s own login
acme-docsHarness’s own loginHarness’s own loginglm

2 agents in acme-shop still run on the previous connection. Restart them to move them to the new one.

Each project’s default connection; agents already running keep theirs until restarted. Sample data.

A key turns it on.

Without a key, Connections and Usage explain what they do and stay off, and agents use the login of the harness on this machine, one per harness. Add a harnsy Max key on the License page to turn them on.

Apply for a key →

What's in harnsy, and what's in harnsy Max? →

Limits, honestly

  • Without harnsy Max, agents use the login of the harness on this machine, one per harness.
  • Signing in to Codex by device code hasn’t been tried live yet.

How to set it up →

Your first team is one prompt away.

Installing takes one sentence to your agent. To run your agents, harnsy drives a terminal: tmux if it is installed, otherwise its own terminal host (on Windows, always its own).

Install harnsy following https://harnsy.dev/llms.txt

Questions

What's in harnsy, and what's in harnsy Max?

harnsy itself is free: 1 organization, 2 projects, no limit on agents, one machine. Claude Code, Codex, OpenCode and pi need their own plans or keys, paid to their providers. harnsy Max is the same build with a key: more projects and organizations, access from your phone (beta), several machines as one team, several subscriptions and their usage.

What will it cost in total?

harnsy adds no AI charges: your agents run on your own AI plan — a Claude subscription, ChatGPT/Codex, or a model you connect through OpenCode — paid on the provider’s site. The free edition of harnsy runs 2 projects in one organization; harnsy Max, the paid edition, adds more projects and organizations. More agents at once use the plan’s limits faster; the Usage page shows how much is used. Their sites show the prices. harnsy doesn’t sell model access.

All questions →
site-3flead · Claude Codeliveview only
❯ Plan #42 with the team.

Waiting for the breakdown from analyst-7a…

from analyst-7a through harnsy❯ #42 broken down: three acceptance criteria, including a retry after 24 h.

I’ll hand #42 to site-9a.

❯