Skip to main content

Buyer guide

T3 Code Team Setup: Agents, Permissions and Remote Access

Decide which agents and accounts T3 Code may drive, set the permission mode, and lock down remote access before a team relies on it.

Reading time
2 min read
Sources
12 cited
In this piece
T3 Code+20 Positive

The call

Roll T3 Code out with the providers, permission mode and Claude account rule chosen first; an engineer who installs it alone gets Full access and every agent their machine can reach.

How should a team start using T3 Code without inheriting its most permissive defaults? Decide which agents and accounts it may drive, set the permission mode, keep remote access and telemetry deliberate, and read each turn's diff before merge.

When you need this

  • Engineers run two or more vendors' agents and want one supported setup.
  • Someone wants to steer agents from a phone or a second machine.
  • Security asks where prompts go and what the app reports back.
  • The team pays for Claude plans and wants to run Claude through T3 Code.

Run the setup

  1. Before work: choose agents and accounts

    Which agents may T3 Code drive, and on whose login?

    What good looks likeA short list of providers per project, set in project settings, with each login named: ChatGPT plan or Codex login, Claude login or API key, a paid Cursor plan, Grok Build, OpenCode, Antigravity. T3 Code requires at least one provider installed and signed in before use.

  2. Before work: settle Claude billing

    Does Claude run on a plan or a key?

    What good looks likeA written decision. T3 Code drives Claude through Anthropic's Agent SDK; Anthropic asks Agent SDK products to use API keys and says third-party app use draws from plan limits.

  3. Before work: pin a stable release

    Which build will engineers run?

    What good looks likeA stable release, not a nightly. T3 Code shipped 7 stable releases and 117 nightly builds in the 30 days to Nov 1, 2026.

  4. During work: set the permission mode

    What may an agent do without asking?

    What good looks likeSupervised or Auto-accept edits for new threads instead of the Full access default, with Full access kept for isolated machines.

  5. During work: use parallel worktrees on purpose

    Where does each model's attempt land?

    What good looks likeSeparate threads and git worktrees per model, with automatic cleanup on in Settings > Storage, since T3 Code warns worktrees add up fast.

  6. Before merge: review each turn's diff

    What did this turn change?

    What good looks likeThe per-turn diff T3 Code shows, then the branch comparison for the full change, before any commit, push or pull request it generates.

Questions for T3 Tools

  1. Will T3 Tools document an admin policy that fixes the permission default for every engineer?
  2. Has Anthropic approved Claude plan logins through T3 Code's Agent SDK route?
  3. What changes between stable releases that a team's settings depend on?
  4. Which telemetry fields leave the machine, and can an administrator turn them off centrally?

Red flags

  • New threads still in Full access on a machine that holds production credentials.
  • Nightly builds on engineers' daily machines.
  • Claude plan logins shared across a team through one T3 Code server.
  • A server reachable from other machines with no decision on who may link to it.

Products in this piece

Sources

Official documentation
10
Public source
2
  1. T3 Code READMEOfficial documentationgithub.com
  2. T3 Code docs: FAQOfficial documentationgithub.com
  3. T3 Code docs: Install and first runOfficial documentationgithub.com
  4. T3 Code docs: Permission modesOfficial documentationgithub.com
  5. T3 Code docs: Project settingsOfficial documentationgithub.com
  6. T3 Code docs: Thread sidebarOfficial documentationgithub.com
  7. T3 Code docs: Source controlOfficial documentationgithub.com
  8. T3 Code docs: Remote accessOfficial documentationgithub.com
  9. T3 Code docs: TelemetryOfficial documentationgithub.com
  10. T3 Code releases on GitHubOfficial documentationgithub.com
  11. Anthropic: Claude Code legal and compliance (read Nov 1, 2026)Public sourcecode.claude.com
  12. Anthropic support: Use the Claude Agent SDK with your Claude plan (read Nov 1, 2026)Public sourcesupport.claude.com

Keep reading

  1. ResearchSeveral Models in One App: Routing, Benchmarks and Native Tool Costs in T3 CodeWhich agent reaches which model in T3 Code, what independent benchmarks say about each kind of coding work, and when a model API's own search or video tool costs less than the data source's API.
  2. ReviewT3 Code Review: One Open-Source Control Surface for the Coding Agents You Already Pay ForWho T3 Code fits, what it adds over each agent's own app, and what a team must settle before it runs Claude, Codex and other agents side by side through it.
  3. Buyer guideAider Workflow PlaybookSix working habits for teams whose Aider results vary from one engineer to the next, each with the question a lead can ask to check it.
Read the research