tethercode.
Get early access
What matters most to you?

Your AI agents don't talk to each other. tethercode changes that. Stop re-explaining your project to every agent. Pay for the expensive model only when the task needs it. Your code, your memory, your rules. Your agents are already inside the company. Can anyone say what they did?

Claude Code, Codex, Antigravity, Pi and your local models message each other and work from one shared context: the same memory, the same task ledger, the same map of your code. You stop pasting background into every chat and carrying news from one agent to the next.

Tell one agent once. The next one already knows what the project is, what was decided and what is half done. Install tethercode once, then keep using Claude Code, Codex, Antigravity or Pi exactly as you do now.

The searching, summarising and research your agents hand to tethercode run through one router. Of the models allowed to see a call and able to do it, it tries the most private first, then the cheapest, and records what each call cost. Your expensive model gets the answer, not the reading.

tethercode runs on your machine. The code map, the agents' memory and the task ledger live in your own database, not ours. What may go to a cloud model and what must stay with a local one is your setting, checked on every single call.

What they read, what they changed, what it cost and who allowed it. tethercode makes every agent act as a named person with that person's access, reads every model call before it leaves, asks a human before anything risky, and records every decision. On your servers and your GPU.

Free for 3 repositories, with no time limit and no card. One email when it is ready, and nothing else.

$0 for 3 repositories. $19.99 a month for everything, with 500 million helper-model tokens a month included. One email when it is ready.

Trying it alone first is free. For a company deployment, talk to us.

Works with Claude Code, Codex, Antigravity, Pi, local models and any provider's key. It is not another AI.

What working with AI agents is like today

The models are good. Everything around them is missing, so you fill the gap by hand.

The same four cords in a hopeless tangle
  1. Every session starts from zero

    You explain the project, the decisions and the half-finished work again. Each tool remembers only itself, if it remembers anything.

  2. Your agents collide

    Two of them edit the same file. One undoes the other's work. Nobody knew who was doing what, and you find out when the tests fail.

  3. They change code they cannot see around

    An agent rewrites a function without knowing what calls it. It works here and breaks something three files away.

  4. They skim, because reading costs tokens

    Every file an agent opens and every search it runs is paid for in tokens. So it reads a few files, searches once and guesses the rest, then forgets half of that when the conversation gets long. Most agent mistakes are things it never read or no longer remembers.

  5. The expensive model does everything

    Including the work a cheap or local model would have done just as well. You find out on the bill.

  6. The safety check is a list of words

    An approval step that looks for "delete" or "restart" is defeated by the first synonym. An agent only has to find the one word you forgot, and it will find it while doing something else.

  7. And nobody can say what the agents did

    What they read, what they changed, what it cost, who allowed it. For one person that is a nuisance. In a company it is the reason the answer is still no.

One install that all your agents share

tethercode is software you install once, on your own computer or your company's servers. The AI agents you already use connect to it and keep working as before. It is not another AI: your agents and models still do the thinking.

What it adds for all of them:

  • One shared memory and task list: no agent starts from zero
  • Messages between agents, even from different vendors
  • A map of your code that answers with file and line
  • Your approval before a risky action, and a check on every model call it makes

Everything it stores is in your own database. Free for 3 repositories, $19.99 a month for all of them.

Why it exists

The same afternoon, with tethercode
  1. Claude Code opened a task: move sessions to the new token format
  2. shared memory decision saved: tokens last 30 minutes, refresh slides
  3. code map Session.refresh has 3 callers. Two are in files Codex is working on.
  4. Claude Code to Codex I'm changing what Session.refresh returns. Your billing tests call it.
  5. Codex to Claude Code Seen. Holding my push until yours lands.
  6. local model opened a task: write the migration notes. Read the decision from memory. Nobody explained it twice.
  7. needs you Codex wants to delete 14 test fixtures. ApproveDenyasked on your phone

Three good reasons, depending on who you are

Someone leaning back from their desk with a mug, relaxed

Because you would rather build than babysit

  • One install on your own computer. A laptop is enough.
  • Keep the agents you already like. Nothing to relearn.
  • Agents pick up where the last one stopped.
  • Risky actions ask you first, on your phone.
A tidy desk seen from above: a notebook of neat figures, a pencil and a calculator

Because the arithmetic works

  • $0 for up to 3 repositories. $19.99 a month for everything.
  • Searching, summarising and research go to the cheapest model that is allowed and able.
  • Every call logged with its cost, so you can check the claim.
  • In our own tests, Claude used about 90% fewer of its tokens on a task when tethercode did the searching.
  • Premium includes 500 million helper-model input tokens a month: $15 of use inside a $19.99 plan.
A small rack of computers at home, with a hand holding an unplugged network cable

Because it is yours

  • Runs on your machine. Your code map, memory and ledger sit in your own database.
  • Works with local models, and lets them talk to each other.
  • You decide, per kind of content, what may ever leave.
  • This page sets no cookies and loads nothing from anyone else. Check the network tab.

What you cannot do without it

Each vendor's tool knows its own agent and nothing else. Everything below needs something underneath all of them. All of it is built, and all of it is in the free edition unless its page says otherwise.

And you can watch all of it

tethercode comes with a console in your browser, served from your own machine. These are the screens people open first, drawn from the console's own.

The agent board: four agents, the task each one has open, what they decided and ruled out, and a request waiting for approval The agent mailbox: a conversation between Claude Code and Codex, with delivery shown on each message and an agreement on record The code map: one function selected, with the three places that must change, the test, the document and four files that reach it indirectly Knowledge: a question answered with three facts and their sources, one thing reported as not found, and one company held as a single identity Models: tokens used this week on your own network and in the cloud, today's calls by model with where each may run, and one call explained step by step Reasoning review: a suspected missing link between two functions with the path through the code that supports it, a rule it has learned, and two records that disagree, each waiting for a decision Research: a finished run with its gaps stated, the report with its numbered sources, and the list of every search and every page fetched, admitted or refused

Every agent, what it is working on, what it has decided and ruled out, and what is waiting for you.

What your agents say to each other, whether each message arrived, and what they agreed. You can write to them too.

Pick any function and see what has to change with it, before anyone edits a line.

Ask a question and get the answer with the records behind it, and a plain "not found" for what it does not hold.

Which model each call went to, why it was allowed to go there, and what it cost.

What it worked out and what it only suspects, each with the path that supports it. Nothing becomes a fact until you confirm it.

A research run with its report, every source it cites, and every search and page behind it, including the pages it refused.

A phone showing two approval requests from agents in a chat, one denied earlier and one with Approve and Deny buttons
And the risky ones reach you wherever you are.

Illustrations of the console's screens, drawn with example data.

Nothing confidential leaves by accident

An agent that has just opened a customer's account should not be sending a summary to a public model thirty seconds later. tethercode reads every model call before it is sent, with a small model called Kev that was trained for that one job. We gave Kev and a general-purpose model the same 1,000 samples.

  • marked public when it was not
  • rated too low in another way
  • rated too carefully
  • right

Each dot is one test sample. The samples are synthetic and the general model was judged on definitions it had never seen: the method, the tables and the limits.

  • 946 against 681 right answers out of 1,000
  • 4 against 193 rated less sensitive than they are
  • 515 of 527 risky agent actions caught before they ran
  • 39 ms to judge a sample, against 418 ms

Stop carrying messages between your agents. Explain your project one last time. Run the numbers on your own repository. Keep it on your machine. Connect it anyway. Start with one team. Keep it inside.

Edition

One email when it is ready. No card now, and choosing Premium here commits you to nothing.

Deploying for a company? Talk to us. What each edition includes. Questions and answers.