The best way to control Claude Code, Codex and more from any computer or phone

main
md 446 lines 25 KB
Rendered Raw
1 # Messages and context
2
3 Give the agent a task in the composer. Add files, quote a previous response, or
4 include a skill when the task needs more context.
5
6 Messages can contain up to 120,000 characters. Longer drafts stay in the composer
7 so you can shorten them or split them into several messages.
8
9 Pasting 32 KiB or more of text adds that fragment as a text-file attachment so
10 the agent can inspect it without filling the model context. A smaller paste also
11 becomes an attachment when inserting it would exceed the message limit. On a
12 hardware keyboard, use `Cmd+Shift+V` on Apple devices or `Ctrl+Shift+V` elsewhere
13 to keep a large paste editable in the composer instead.
14
15 ## Formatting
16
17 The composer writes Markdown and shows it styled as you type. When your cursor
18 is on styled text, its Markdown shows instead, such as `**bold**`, and the
19 markers edit like any other text: delete them to remove the style, or change
20 `**` to `*` for italic. The style updates when your cursor leaves. To write
21 plain Markdown with no styling, turn off **Settings → Chat → Rich text
22 composer**. Lines starting with `- `, `1. `, `- [ ] `, `> `, `# ` or `---`
23 become lists, task lists, quotes, headings and rules; ` ``` ` followed by Enter
24 opens a code block. What you typed is what the agent receives, markers and
25 numbering included, and `#1234` without a space still looks up a pull request.
26
27 Some styled text containing or touching literal Markdown characters, such as a
28 bold `path\` or `2*3`, keeps its styling without showing markers. Use formatting
29 shortcuts or turn off rich text to edit the formatting of those ranges.
30
31 Enter sends. In a list or quote, **Shift+Enter** continues it, and Shift+Enter
32 on an empty line moves out one level, then leaves it; **Tab** nests a list item
33 under the one above and **Shift+Tab** moves it back out. In a code block, Enter
34 starts a new line at the current indentation, **Tab** and **Shift+Tab** indent
35 the selected lines, and a closing ` ``` ` followed by Enter, or two blank lines
36 at the end, leave the block. **Backspace** at the start of a code block turns it
37 back into plain lines. Choose the language in a code block's corner to change
38 it. Very large code blocks are shown without syntax highlighting.
39
40 ## Attach files
41
42 Attach up to 100 files per message. Each image can be up to 10 MiB, with at most
43 80 MiB of images in one message. Other files, including videos, can be up to
44 50 MiB each, subject to the environment's upload support and limit. The agent
45 receives them on the environment's machine. Provider and model limits still
46 apply, including images already in the conversation. A video attachment gives
47 the agent a file path; only [Antigravity](./providers-antigravity.md#skills-and-attachments)
48 sends supported videos to Gemini models as video.
49
50 Uploads begin when you add an attachment. All uploads must finish before the
51 message can send. Retry or remove a failed upload. On web and desktop, reloading
52 before an upload finishes requires you to attach that file again.
53
54 You can drag or paste images into the web or desktop composer. HEIC and HEIF
55 photos are converted to JPEG there and when selected from the mobile photo
56 library; photos over the image limit are also resized to fit. On mobile, you can
57 also send files to T3 Code through another app's system share sheet.
58 On Android, you can also drag images from another app into the message input
59 in split screen.
60
61 See [images and videos](#images-and-videos-in-messages) for previewing and saving media.
62
63 ## Send while the agent is working
64
65 On web and desktop, choose **Settings → Chat → Follow-up behavior** to queue
66 new messages for a later turn or steer the running turn immediately. The setting
67 applies to this client; already queued messages keep their place. Queued messages
68 are saved on the server and can be edited or removed before they send.
69 When Enter sends, `Cmd+Enter` on macOS or `Ctrl+Enter` on Windows and Linux uses
70 the opposite action: it steers when your default is Queue and queues when your
71 default is Steer.
72 If your **Send shortcut** requires `Cmd+Enter` or `Ctrl+Enter`, that shortcut uses
73 your default follow-up behavior; add **Shift** to use the opposite action. This
74 applies only to multiline messages when you choose the multiline send option.
75
76 Use `Cmd+Shift+Enter` on macOS or `Ctrl+Shift+Enter` on Windows and Linux to send
77 the oldest queued message as a steer. This leaves the current draft intact and
78 requires an active turn that supports steering. If that shortcut sends the opposite
79 action for your draft, move focus outside the composer to steer the queued message. Change
80 `thread.steerQueuedMessage` in **Settings → Keybindings** to use another shortcut. To make
81 `Cmd+Enter` or `Ctrl+Enter` steer the oldest queued message when the composer is empty, add
82 this rule to your [keybindings file](keybindings.md#edit-the-configuration-file):
83
84 ```json
85 {
86 "key": "mod+enter",
87 "command": "thread.steerQueuedMessage",
88 "when": "composerFocus && turnRunning && composerEmpty"
89 }
90 ```
91
92 Press `Option+Up` on macOS or `Alt+Up` on Windows and Linux with the cursor at the
93 start of the composer to edit the most recently queued message. Change
94 `thread.editQueuedMessage` to use another shortcut.
95
96 Mobile has the same choice under **Settings → Chat**. While a turn is
97 running the send button shows which action it will take. Long-press it to use the
98 other action for a single message. From a hardware keyboard, `Cmd+Return`
99 (`Ctrl+Return` on Android) uses the other action, or `Cmd+Shift+Return`
100 (`Ctrl+Shift+Return`) when Return inserts a new line. The button only offers
101 Steer when the running agent supports it.
102
103 ## Queue messages offline on mobile
104
105 Mobile keeps local copies of draft attachments, so you can preview them and queue
106 messages while disconnected. Uploads resume when you reconnect. Drafts and queued
107 messages survive app restarts. Signing out of T3 Connect keeps that work on your
108 device until you sign back into the same account.
109
110 ## Custom models
111
112 On web and desktop, use Settings → Providers → **Models** to add an unlisted model with a custom
113 name and options. Only options supported by the provider integration affect turns. Antigravity
114 uses its account catalog and does not support custom models.
115
116 To rename a built-in model on every client, edit the server's `settings.json` in its `userdata`
117 directory. Add a custom model entry with the built-in model ID and a new name under that provider
118 instance:
119
120 ```json
121 {
122 "providerInstances": {
123 "codex": {
124 "driver": "codex",
125 "config": { "customModels": [{ "slug": "gpt-5.5", "name": "Team GPT" }] }
126 }
127 }
128 }
129 ```
130
131 ## Model defaults
132
133 T3 Code remembers your provider, model, and model options for new threads. A
134 project's configured model takes precedence; resetting that project setting
135 returns to the remembered selection.
136
137 Leaving reasoning level or service tier unset uses the provider's own configuration.
138
139 ## Quote an assistant response
140
141 On web and desktop, select text within one assistant response and choose
142 **Cite in composer**. You can add a comment about the quote and write instructions
143 around it.
144
145 Select the quote in a draft or sent message to return to its source. If the source
146 is unavailable or has changed, the saved quote remains readable.
147
148 The chip shows your comment when it has one, or a short quote preview otherwise. Use the pencil
149 button to add or change the comment. To remove the citation, place the caret beside its chip and
150 delete it like other inline context. Copying, reloading, and restoring a
151 [stashed prompt](#prompt-stash) keep each comment
152 with its quote, and sending tells the agent which words were quoted and which comment you wrote.
153 The quoted text and comment count toward the message limit.
154
155 On mobile, long-press an assistant response, adjust the text selection, and choose
156 **Cite** to insert a quote at the composer cursor. If the passage appears more than
157 once in that section, choose its source occurrence. Select more surrounding text
158 if a short passage appears too often. Tap a quote in a draft to add
159 or edit its comment, or to remove it. To open the source message, tap a quote in
160 a draft or sent message, then choose **Go to message**.
161
162 ## Recall a sent prompt
163
164 Press `ArrowUp` in an empty composer to bring back the last prompt you sent in this thread. Press
165 `ArrowUp` again to go further back, and `ArrowDown` to come forward. Moving forward past the newest
166 prompt clears the composer. Recall walks the prompts loaded in the thread. Attachments, terminal
167 context, and other extras from the original message are not restored, only the text you typed. A
168 composer that holds an attachment or a picked element does not count as empty.
169
170 When the composer has text, the arrow keys move the caret as usual. Recall takes over only while
171 the text is an unedited recalled prompt, with the caret on the first visual line for `ArrowUp` or
172 the last visual line for `ArrowDown`, counting wrapped lines. Editing a recalled prompt turns it
173 into a normal draft.
174
175 ## Suggested next prompt
176
177 Turn on **Suggest next prompt** in Settings > Labs. After a Claude or Codex turn completes, the
178 empty composer shows a predicted next message. Press `Tab` to fill it in, then edit or send it, or
179 press `Escape` to dismiss it. Each suggestion is one extra request that reuses the turn's prompt
180 cache, so it uses a little of your plan's usage. Claude skips suggestions on the first turn of a
181 thread and near usage limits. Codex needs version 0.165 or later; if T3 Code manages
182 Codex, update it in Settings > Providers. Codex follows a change to the setting at once; open Claude
183 threads follow it after they restart. Suggestions show on web and desktop.
184
185 ## Edit an earlier prompt
186
187 On web and desktop, choose **Edit from here** beneath a sent message to rewind
188 the conversation to before that message. Choose **Revert and keep changes** to
189 leave workspace files as they are, or **Revert files too** to put the workspace back
190 to its checkpoint from before that message. That also undoes edits you made yourself.
191 File restore is only offered for threads running in a worktree, and it is
192 refused when another thread or agent session also uses that directory, a folder
193 inside it, or a folder that contains it, since
194 restoring would erase their changes. Completed subagents belonging to the reverting
195 thread do not block restoring that same worktree, provided they have no queued or
196 active work, pending requests, or background tasks. A thread that works in the project
197 directory rewinds the conversation only. The selected prompt and its attachments return to the composer for editing and
198 resending. Any unsent draft stays above the restored prompt.
199
200 This removes the selected message and later conversation from the active thread
201 and provider history. It does not undo external actions or separate provider
202 memory. The action is available only when the provider supports rewind.
203
204 ## Prompt stash
205
206 On web and desktop, press `Cmd+S` on macOS or `Ctrl+S` on Windows and Linux to save
207 the current prompt and its attachments for later. Wait for uploads to finish first.
208 With an empty composer, the same shortcut restores a single stash or opens the
209 stash menu when there are several.
210
211 Stashes containing uploaded files must be restored in their original environment.
212 Those files are retained for 24 hours. After an upload expires, restore the prompt
213 and use **Attach again** or remove the missing file before sending.
214
215 ## Voice input on web and desktop
216
217 Use the composer's microphone or `mod+shift+d` to dictate into your draft. Audio
218 is transcribed on an environment you choose in **Settings → Voice**, or on the Mac
219 with Apple Speech in the macOS desktop app; nothing is sent until you choose one.
220 See [Voice input](./voice-input.md).
221
222 ## Voice input on mobile
223
224 Use the composer's microphone to record, then confirm to transcribe. Text is
225 inserted where your selection was when recording started, ready for you to review
226 and edit before sending.
227
228 By default, audio stays on the phone. Local transcription needs a supported iPhone
229 with iOS 26 or later, or Android 12 or later with compatible Google speech services.
230 To use a paired environment instead, open **Settings → Voice → Transcribe on**,
231 choose the environment, and confirm. Finished recordings go only to that environment,
232 even when a draft belongs to another one. This also works on phones without local
233 speech recognition. If the chosen environment disconnects, dictation stops; it never
234 switches hosts. Choose **This device** to revoke access and discard pending audio.
235 For [cloud dictation](./voice-input.md#choose-where-audio-goes), open **Cloud providers**
236 under the environment in **Settings → Voice** and confirm which provider may receive recordings.
237 On Android, keep T3 Code open while recording and transcribing.
238
239 On-device dictation may need a network connection to download a speech model the
240 first time. After that it works offline for that language. iPhone dictates in the
241 first of your preferred languages (**Settings → General → Language & Region**)
242 that Apple’s speech model supports. If none is supported, it tries the app’s
243 language. Android transcribes on the device while you
244 record and finishes after you stop. Recordings can be up to five
245 minutes long. Canceling or an audio interruption discards the recording and
246 preserves your existing draft. Navigating to another screen keeps dictation active;
247 use the floating controls to finish or cancel. While recording, the screen stays
248 awake; it can sleep normally once recording stops.
249
250 Choose a preferred order in **Settings → Voice**. Voice input requests the first
251 connected microphone in the list. The list mixes kinds (built-in, other wired, other
252 Bluetooth, and CarPlay on iOS) with wired and Bluetooth devices you have dictated
253 with, listed by name. Drag to rank them together:
254 for example, put **Other Bluetooth** above your AirPods so a new headset is preferred
255 while the AirPods microphone stays low. By default, CarPlay comes after the built-in
256 microphone on iOS. Swipe a device left to forget it. Android may use another input
257 when the preferred microphone cannot be routed. Devices with the same name may share
258 a saved preference, but simultaneous same-name inputs remain separately selectable.
259
260 T3 Code deletes the temporary recording from your phone after transcription or
261 cancellation. The transcript stays in your draft until you send the message to the agent.
262
263 ## Queued messages
264
265 On web and desktop, the composer shows **Interrupt** while the agent is working and the draft is
266 empty. Adding text or attachments replaces it with a send button that shows your
267 [Follow-up behavior](#send-while-the-agent-is-working): a queue icon for Queue or a steer arrow for
268 Steer. Click it or press your send shortcut to use that action. Hold `Cmd` on macOS or `Ctrl` on
269 Windows and Linux to switch the button to the opposite action, then click, or use the opposite-action
270 shortcut described above.
271
272 On web and desktop, queued messages appear at the end of the thread as dashed messages that have
273 not been sent yet. On mobile they appear above the composer. Each shows its text and any attached
274 images. **Send now** sends a message right away, into the running turn if there is one. You can
275 also edit or remove it.
276
277 If you stop the agent or the server restarts, queued messages keep their order and are held.
278 Press **Resume** in an empty composer to continue. For stopped runs and usage limits,
279 this uses the provider's native continuation without sending a new prompt. If the provider
280 does not support it, send a message to continue instead. If the original request never reached
281 the provider, Resume sends that request again.
282
283 Sending a new message also resumes the queue, unless it steers a turn that is already
284 running. You can edit or remove held messages without starting them.
285
286 The pencil on a queued message opens it in the composer for editing. The original message
287 stays in the queue until you save, and it is highlighted while you edit. The message's
288 attachments appear above the text with a remove control, and new images can be added the usual way.
289 The checkmark saves the queued message in place; **Cancel** next to it leaves it unchanged. Whatever
290 you had typed in the composer before starting the edit is restored afterwards. If the queued
291 message starts or is removed while you are editing, the edit ends: changed content moves into the
292 composer when it is empty, and is discarded otherwise.
293
294 ## Commands and skills
295
296 Type `/` for commands or `$` to add a skill from the selected environment and
297 provider. On mobile, both are also available before starting a thread on
298 **New task**.
299
300 The slash menu also includes skills unless you turn off **Settings → Chat →
301 Show skills in slash menu**, under Advanced. Only skills enabled for the provider are listed.
302
303 After you add or change skills, plugins, or MCP servers, use **Restart agent
304 session** in the command palette on web and desktop. The conversation continues,
305 and your next message starts the agent again with the new setup.
306
307 Provider commands must start the message to run. T3 Code commands such as
308 `/model` and skill mentions work on any line.
309
310 Send `/compact` in an existing conversation to reduce context usage when the
311 provider supports it.
312
313 Send `/clear` on its own, or pick it from the slash menu, to start a fresh session
314 in the same thread. Your next message goes to a new provider session that doesn't see the
315 earlier messages, and the thread marks where it began with **New session**.
316 Earlier messages and diffs stay in the thread. Dismiss the notice above the
317 composer to keep the current session instead.
318
319 On web and desktop, turn on **Context window** in **Settings → Appearance →
320 Thread details** to see context usage in the thread details card. It appears
321 once the provider reports usage.
322
323 Claude instances can also compact idle threads for you. See
324 [compact long conversations](./providers-claude.md#compact-long-conversations).
325
326 ## Goals
327
328 With Codex, Claude, and OpenCode 2.2.5 or later, send `/goal` followed by what "done" means, for example
329 `/goal all tests in packages/api pass`. The agent keeps working across turns
330 until it judges the goal met. The thread shows **Goal** while it works, and a
331 line at the end of the thread (above the composer on mobile) shows the goal and
332 its progress.
333
334 - `/goal` alone shows the current goal. `/goal clear` removes it.
335 - Codex and OpenCode also support `/goal pause` and `/goal resume`. Stopping
336 their goal pauses it. Codex also pauses when its work ends without meeting the goal.
337 - Stopping Claude ends the current turn, but the goal stays set. Claude checks it
338 again at the end of your next message.
339 - Sending a new goal, or `/goal resume`, while the agent works stops the current
340 turn and starts the goal, unless you queue it. `/goal clear` and `/goal pause`
341 stop the turn only when it is working on the goal. Other goal commands wait for
342 the turn to end.
343
344 ## Context in your message
345
346 Context you attach lands where your cursor is, as a chip inside your text: a terminal excerpt,
347 a review comment from a diff or file, a preview annotation, or a file. You can type before and
348 after a chip, move it by cutting and pasting, and delete it like a character. Hover a chip for
349 its brief details. Select a terminal excerpt to open its captured output, or select a review
350 comment, picked element, or preview annotation to open its full details. Chips read as "Terminal
351 excerpt, Terminal 1 lines 3-4" and similar to screen readers.
352
353 A pull request appears as its icon and number. Its color reflects whether it was open, draft,
354 merged, or closed when it was attached. Select it to inspect the captured title and branches,
355 then choose **Open pull request** to visit the pull request. On web and desktop, type `#` to browse the newest
356 pull requests in the current project's repository. Pull requests linked to the thread are included
357 from any repository and listed first, after any exact number match. Continue typing digits to filter the recent list
358 by any part of its pull request numbers. A complete number is also resolved directly, even when that
359 pull request is older than the recent list. Type a single word after `#` to search pull requests in
360 the repository by text. Choose a result to insert it as a chip.
361
362 Another thread can be context too. Type `@` followed by part of its title to pick one from
363 the same server, or from a machine it links to while the Linked environments beta is on; those
364 show the machine's name. On web and desktop you can also drag a thread out of the sidebar and
365 drop it on the composer; a multi-selection drops together. On any client, **Copy thread** copies a
366 link to the thread that pastes into a composer as its chip. The chip shows the thread's
367 current title and opens it when selected. Your prompt only carries a reference: the agent reads the thread's
368 history on demand, so attaching a long thread costs nothing until the agent looks. Attaching a
369 thread does not change it, and the agent cannot send messages to it unless you ask.
370
371 A thread from a linked machine, picked, pasted or dropped, stays a reference that the agent reads
372 on that machine. A pasted link to a thread on any other machine stays a link. Dropping a thread
373 from any other environment attaches a saved transcript file. The source
374 environment must be connected and up to date. The transcript includes the visible history and
375 saved text context at the time of the drop; later messages and the contents of attached files
376 are not copied. The agent can read the saved transcript after the source disconnects. Normal
377 file attachment limits apply.
378
379 Type `@` and part of a model or provider name, such as `@opus`, to name that model for the
380 agent, for example when delegating. It does not change the thread's model. Join words with
381 `:`, `/` or `-`, as in `@opus:opencode`.
382
383 Images keep their thumbnail shelf above the text and also get a chip at your cursor, so you can
384 say exactly which image you mean. Deleting an image chip leaves the image on the shelf; removing
385 the thumbnail asks first when the image is still mentioned in your text, then removes both. Files
386 exist only as chips: deleting a file's last chip removes the file from the message.
387
388 Copy text that holds chips and paste it into another draft, in the same thread or another one,
389 and the chips come along with what they point to. Images and files are fetched again from the
390 environment they came from; while that happens the chip shows a dashed outline, and if it cannot
391 complete T3 Code tells you and leaves the chip for you to remove or replace. A chip whose
392 context is no longer available shows the same dashed outline; hover it for what to do.
393
394 Copying a message with the copy button, or copying text out of it, gives other apps readable
395 Markdown with a link in place of each chip. Older messages that were sent before chips still
396 show their context. Stashing a prompt keeps its chips and what they point to; restoring brings
397 them back.
398
399 On mobile, tap a chip to inspect its content. File references open the current file; attached
400 files show the copy that was attached to the message.
401
402 ## Attached files
403
404 Select a file chip in your draft or a sent message to preview it. Code and JSON use syntax
405 highlighting; Markdown, HTML, CSV, and TSV offer rendered and raw views. Audio files have
406 playback controls. Large text files show a limited preview; save the file to read it in full.
407
408 On web and desktop, files open beside the conversation with the same controls as a workspace
409 file: a header row with the view toggle, **Copy contents** and **Save file**. On mobile, documents
410 open in the same file screen as workspace files; its menu holds **Copy contents**, **Save or
411 share** and **Open in file viewer**. Pictures, videos and PDFs keep their native viewers, and
412 other document formats such as Word or Pages open in the device's own viewer when it has one.
413 If nothing on the device can show a format, save or share it to open it elsewhere. A ZIP up to
414 50 MB opens as a folder list on every surface; on mobile, select a file inside it to preview,
415 save, or add it to chat.
416
417 ## Images and videos in messages
418
419 Select an image or video attachment or link to preview it. Playback support depends
420 on your browser or device; save an unsupported video to open it in another app.
421
422 On web and desktop, right-click media to save it or copy its path or URL. On mobile,
423 touch and hold an image or video thumbnail and choose **Save or share**. On iOS,
424 return to the thumbnail to open this menu after watching a full-screen video.
425
426 File links refer to the environment's machine, including when you connect remotely.
427 Previews use the original file, even outside the workspace. Moving or deleting it
428 can break the preview, so save a copy if you need to keep it.
429
430 ## Files outside the workspace
431
432 Follow an agent's file link to read a report or other file outside the workspace.
433 These files open read-only. An HTML file outside the workspace cannot load scripts,
434 styles, or images from neighboring files.
435
436 ## HTML and PDF files in the file viewer
437
438 On web and desktop, HTML and PDF files open as rendered pages. Switch an HTML
439 file to source view to read its markup; a link to a specific line opens source
440 automatically. HTML previews cannot access your T3 Code session.
441
442 The file viewer recognizes images, HTML, and PDF files by their filename extension,
443 including filenames or folders containing `#` or `?`.
444
445 On mobile, select a PDF attachment or link to open it. iOS uses the native viewer;
446 Android opens a compatible installed file viewer.