This tutorial of T3 Code focuses on the specific task in the title. T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
Below is a demonstration using a test environment and revocable data. When accounts, keys, machine control, investments or emergency communications are involved, manual confirmation must be done.
T3 Code manages sessions instead of model accounts
This section only deals with “T3 Code manages sessions instead of model accounts”. Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
Check the Node.js version on Windows first
It is recommended to execute in the following order:
- Check the actual version and source.
- Use a test account, test directory, or test device.
- Perform an undoable action.
- Save the output, exit code, or screenshot.
- Revert the changes and re-verify.
|
|
The completion standard is to “check the Node.js version on Windows first” to get verifiable results, rather than the command just not reporting an error.
Obtain the source code from the official repository
| Check items | Passing criteria | Stop signal |
|---|---|---|
| Get the source code from the official repository | Clear input, output and permission scope | The tool automatically expands the scope of operations |
| Source | Can return to the official warehouse or original data | Rely on unknown mirrors or second-hand conclusions |
| Security | No secret value in log | Token, cookie or private data present |
| Recovery | Ability to return to pre-execution state | No backup or undo methods |
|
|
When a stop signal appears, resume first and do not continue to the next step.
Check the package manager lock file before installing dependencies
This section only deals with “checking the package manager lock file before installing dependencies”. Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
Confirm the listening address after starting the Web UI
It is recommended to execute in the following order:
- Check the actual version and source.
- Use a test account, test directory, or test device.
- Perform an undoable action.
- Save the output, exit code, or screenshot.
- Revert the changes and re-verify.
|
|
The completion standard is to “confirm the listening address after starting the Web UI” to obtain a verifiable result, rather than the command just not reporting an error.
Check PATH inheritance when Codex not found
| Check items | Passing criteria | Stop signal |
|---|---|---|
| Check PATH inheritance when Codex is not found | Clear input, output and permission ranges | Tools automatically expand the scope of operations |
| Source | Can return to the official warehouse or original data | Rely on unknown mirrors or second-hand conclusions |
| Security | No secret value in log | Token, cookie or private data present |
| Recovery | Ability to return to pre-execution state | No backup or undo methods |
|
|
When a stop signal appears, resume first and do not continue to the next step.
Claude is logged in but T3 Code is not recognized
This section only deals with “Claude is logged in but T3 Code is not recognized”. Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
How to authorize multiple project directories separately
It is recommended to execute in the following order:
- Check the actual version and source.
- Use a test account, test directory, or test device.
- Perform an undoable action.
- Save the output, exit code, or screenshot.
- Revert the changes and re-verify.
|
|
The completion standard is “how to authorize multiple project directories separately” to obtain verifiable results, rather than the command just not reporting an error.
How does the session list correspond to the real process?
| Check items | Passing criteria | Stop signal |
|---|---|---|
| How the session list corresponds to the real process | Input, output and permission scope are clear | The tool automatically expands the scope of operation |
| Source | Can return to the official warehouse or original data | Rely on unknown mirrors or second-hand conclusions |
| Security | No secret value in log | Token, cookie or private data present |
| Recovery | Ability to return to pre-execution state | No backup or undo methods |
|
|
When a stop signal appears, resume first and do not continue to the next step.
What to check first if the browser page cannot be opened?
This section only deals with “What to check first if the browser page cannot be opened”. Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
Why can’t remote access directly open the port?
It is recommended to execute in the following order:
- Check the actual version and source.
- Use a test account, test directory, or test device.
- Perform an undoable action.
- Save the output, exit code, or screenshot.
- Revert the changes and re-verify.
|
|
The completion standard is “Why can’t remote access directly open the port” to obtain a verifiable result, rather than the command just not reporting an error.
Solve application authentication before reverse proxy
| Check items | Passing criteria | Stop signal |
|---|---|---|
| Solve application authentication before reverse proxying | Clear input, output and permission ranges | Tools automatically expand the scope of operations |
| Source | Can return to the official warehouse or original data | Rely on unknown mirrors or second-hand conclusions |
| Security | No secret value in log | Token, cookie or private data present |
| Recovery | Ability to return to pre-execution state | No backup or undo methods |
|
|
When a stop signal appears, resume first and do not continue to the next step.
Agent accepts files from Git after modifying them
This section only deals with “acceptance from Git after the Agent modifies the file”. Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
Ending the session is not equal to ending the child process
It is recommended to execute in the following order:
- Check the actual version and source.
- Use a test account, test directory, or test device.
- Perform an undoable action.
- Save the output, exit code, or screenshot.
- Revert the changes and re-verify.
|
|
The completion criterion is that “ending the session is not equal to ending the child process” and getting a verifiable result, rather than the command just not reporting an error.
What to keep when upgrading T3 Code
| Check items | Passing criteria | Stop signal |
|---|---|---|
| What to keep when upgrading T3 Code | Clear input, output and permission scope | Tool automatically expands the scope of operations |
| Source | Can return to the official warehouse or original data | Rely on unknown mirrors or second-hand conclusions |
| Security | No secret value in log | Token, cookie or private data present |
| Recovery | Ability to return to pre-execution state | No backup or undo methods |
|
|
When a stop signal appears, resume first and do not continue to the next step.
Restore the original CLI workflow after uninstallation
This section only deals with “restoring the original CLI workflow after uninstallation”. Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
The working directories of different Agents cannot be mixed.
It is recommended to execute in the following order:
- Check the actual version and source.
- Use a test account, test directory, or test device.
- Perform an undoable action.
- Save the output, exit code, or screenshot.
- Revert the changes and re-verify.
|
|
The completion standard is “the working directories of different Agents cannot be mixed” to obtain verifiable results, rather than the command just not reporting an error.
The session is still there after the Web UI is refreshed
| Check items | Passing criteria | Stop signal |
|---|---|---|
| The session is still there after the Web UI is refreshed | Input, output and permission scopes are clear | Tools automatically expand the scope of operations |
| Source | Can return to the official warehouse or original data | Rely on unknown mirrors or second-hand conclusions |
| Security | No secret value in log | Token, cookie or private data present |
| Recovery | Ability to return to pre-execution state | No backup or undo methods |
|
|
When a stop signal appears, resume first and do not continue to the next step.
Where should the log be written when the background is started?
This section only deals with “where the log should be written when the background is started.” Save the current state first and then run the minimal test. Do not change multiple conditions at the same time.
|
|
For T3 Code, results must be judged within this boundary: T3 Code is a web interface for managing multiple coding agents; it does not replace Codex, Claude, Cursor, or OpenCode’s own installation, login, and permissions models.
If the results are not repeatable, record the version, input, and logs, then narrow the scope of the test.
Final check of T3 Code
- The versions and sources used are documented.
- Test input and formal data have been isolated.
- Commands, configurations and actual output can correspond.
- The failed path was actively tested at least once.
- Keys, accounts and private data do not enter Git or logs.
- There are clear actions for upgrading, stopping and resuming.
If you still can’t explain how the results are produced, continue to keep T3 Code in the test environment without expanding the permissions or replacing the existing stable process.