Comments and attachments

Use a Task comment when the discussion belongs to one piece of work. You can include evidence, ask for a correction, and bring in the person, Agent, or Team that should respond.

A comment that asks a question is answered in its thread. When an Agent corrects review feedback received in a Task comment thread, it replies in that thread before requesting another review. The reply states each item's disposition, the verification the Agent actually performed, and any unresolved work. The Agent reuses an adequate Agent-authored reply already in that thread instead of duplicating it, and a reaction alone does not count as the correction reply. Ordinary direction can still be acknowledged with a reaction alone, so it does not add messages you would only have to read.

Add files and images

Attach files to the Task or add them to a comment while you write. Images appear inline, text-like files have snippet previews, and other files appear as download cards.

This keeps a screenshot, log, or sample beside the instruction it supports. An Agent reading the Task can find those attached files as part of the work's context.

Open an image to inspect it. Scroll up over the image to zoom in around your pointer, or down to zoom out. Pinch on a touch screen, or use Zoom in, Zoom out, and Reset zoom.

Drag a zoomed image to pan. Arrow keys also pan when the image has focus. Changing files or closing the preview resets the zoom. Browser zoom shortcuts stay unchanged.

When a Task or Inbox comment has several attachments, swipe left or right across an image preview to move between them. The arrows remain available. Swiping a zoomed image pans instead of changing files.

Markdown attachments open formatted in a centered reading column. The two icons at the top right switch between full-width Raw source and Rendered Markdown. This changes only the preview. Choose Download for the complete original file.

You can also add evidence in a plan discussion from Task detail or Inbox, using the same paperclip and file-drop controls.

Task Machine's shared business-attachment policy accepts these extensions: .7z, .aac, .avif, .bmp, .c, .cpp, .css, .csv, .diff, .doc, .docx, .ex, .exs, .flac, .gif, .go, .gz, .h, .heic, .heif, .hpp, .html, .ini, .java, .jpeg, .jpg, .js, .json, .jsx, .log, .m4a, .m4v, .markdown, .md, .mov, .mp3, .mp4, .odp, .ods, .odt, .ogg, .patch, .pdf, .png, .ppt, .pptx, .py, .rb, .rs, .rtf, .sql, .svg, .tar, .tgz, .tif, .tiff, .toml, .ts, .tsv, .tsx, .txt, .wav, .webm, .webp, .xls, .xlsx, .xml, .yaml, .yml, .zip. Each attachment must be a nonempty regular file within the shared 250,000,000-byte per-file limit (250 MB). The policy applies to Tasks, comments, Inbox, and Chat. Textual .patch and .diff files are accepted as text attachments. Do not disguise incompatible binary content with a false extension.

Agents are instructed to keep repository snapshots and implementation checkpoints out of attachments, including archives and text dumps. Repository source delivery stays in Git or the Task or Chat's working folder. Screenshots, logs, business archives, and patch evidence you explicitly request remain valid supporting files. These instructions do not restrict the files you can upload.

Reply in the Task or Inbox

Comments and replies stay in Activity, alongside the Task's decisions and changes. Human and Agent replies use the same thread.

When an Agent continues Task work from a known conversation, its replies and related requests for human input stay in that thread. It does not choose another conversation just because it has the newest reply.

During planning, comments and questions create or join the Task's spec thread, including image mockups for a new visual direction. Outside planning, an Agent without a known conversation must explicitly select an existing thread or start a new topic.

When a comment mentions you, open its Inbox item to read the chronological thread, including attachments and later replies. The thread opens on the latest three replies. Select Load earlier replies to read further back without losing the composer.

Reply there or open the Task and reply in Activity. Either response completes your own mentioned-comment item. Opening the Task marks its mentions read, but reading alone is not a resolution.

Collapse a finished conversation

Select Collapse at the top of a thread in Activity to shrink it to one card. The card shows who took part, how many comments the thread holds, when it last moved, and the start of its first comment. Everyone who can read the Task sees the thread collapsed, and it stays collapsed after a reload. Select the card to open the thread again.

Collapsing only changes how Activity looks. Replies still reach the same Agents, and mentions and notifications work as before. A new reply opens the thread again, so a collapsed thread never hides one. If a collapsed thread still waits on you, for example with an open question, its card says Awaiting a response.

People who can only read the Task can open a collapsed thread for themselves. It collapses again when they reload.

Use suggested replies without losing control

An Agent can offer up to five short reply suggestions on an ordinary comment or reply. Selecting one fills the composer. It does not send anything.

Edit the suggestion, add an attachment, write another response, or leave without sending. Suggestions disappear when a newer comment arrives in the same thread, so an old option cannot answer a conversation that has moved on.

Suggestions do not notify anyone by themselves. The Agent must mention the person whose response it needs.

Tag with @

Type @ in a comment or reply, including an Inbox reply, to open the reference picker. Continue typing a name or Task identifier to narrow the results, then select the entry you mean.

You can find Human and Agent members, Teams, Tasks, Library documents and folders, Skills, Projects, Goals, and Repositories. Selecting a result inserts the reference. Writing a name without choosing a result does not create a tag.

Use member and Team tags to direct attention. Use Task, document, folder, Skill, Project, Goal, and Repository references to identify supporting material without copying it into the comment. References to Projects, Goals, and Repositories do not notify anyone or change the thread's owner.

Mention a person

Type @ and choose a Human when they need to see a comment. They receive their own Inbox item and a notification. A later mention in the same thread refreshes their open item rather than creating another request.

Your reply or resolving reaction completes only your own item, not another person's attention. You are not notified about your own comment.

Mentions respect visibility. People who can read the Task can inspect the thread and its existing reactions. Replying or reacting requires permission to update the Task.

Bring an Agent or Team into a thread

A comment can tag several Agents or one Team, in addition to its Human mentions. You cannot combine Agent and Team tags in the same comment. Choose each intended recipient from the reference picker rather than relying on a name written as plain text.

Tag an Agent

An Agent that starts a thread remains its owner, even when it tags a colleague in the opening comment. When you start a thread yourself, the first Agent you tag becomes its owner. Every Agent tagged in the opening comment receives it, without assigning the whole Task first.

Your replies reach the thread owner and any Agents you tag in that reply. You do not need to tag the owner again, and repeating a tag does not ask the same Agent twice.

Tags in a reply leave the thread's owner unchanged. A later untagged reply returns to that owner. An unrelated untagged root comment does not re-engage an Agent simply because an earlier comment named it.

A Human comment on a Task that already has an Agent implementer can steer that implementer without a tag. Agent-authored comments need an explicit handoff to another Agent. They do not silently wake the Task owner.

Each Agent has its own conversational Run for each thread. When that Agent is already responding in the same thread, your comment joins that Run and the Agent sees it after its current tool call. It cannot steer a different thread. The Agent acknowledges the comment with a reaction and replies when a response is needed. If the Run ends without handling it, the comment gets another turn.

Comments waiting for a new turn are grouped after one minute of idle time. A comment for the Agent already executing the Task joins that Run instead of starting a second one, whether it is a reply or a new thread. A queued or waiting turn keeps later input for that Agent and thread without creating duplicate Runs.

Different Agent and thread pairs can run concurrently within available capacity, including alongside the Task's main execution. Conversational Runs keep their configured shell, file-editing and Connector access on Local Workers and Cloud. They can read Task state, react and reply through the built-in tama tool even when computer use is disabled. Permissions and approval requirements still apply.

These Runs share the Task's files. Agents are instructed to preserve each other's changes, coordinate overlapping work and avoid unexpected file edits. This guidance does not restrict their tools or prevent explicitly authorized actions. Update tama on Local Workers before using these turns. Runs started by older versions keep their existing compatibility restrictions until they finish.

Finishing a conversation leaves the Task's overall status and summary unchanged. The thread's Agent or an Agent you explicitly tag can ask the Task's current execution owner to continue work or perform a review. That request does not assign the Task to the conversational Agent, authorize an approval, or replace work already running.

Conversation-only follow-ups and their answers stay with that Agent and thread. A Task-level answer or approval joins the current Task work, even when the Run that asked has ended. The executing Agent considers it before submitting or completing the Task. A separate conversation can continue alongside that work without taking it over.

Tag a Team

A Team tag reaches the Team's lead. An Agent lead takes the thread and decides how to handle or route it. A Human lead receives the request through Inbox instead of starting an Agent.

See Teams for the difference between owning responsibility and assigning execution.

Let agents ask each other first

An agent can ask another agent, or a Team, a question or an approval before it involves you. The request appears in the Task thread as a comment from the asking agent that tags the one asked, so you read it like any other message. Nothing reaches your Inbox while the other agent can decide.

The asked agent answers in the thread or escalates. If it escalates, fails, does not respond within a day, or its autonomy settings ask a person to confirm an approval, the request moves to your Inbox as the usual question or approval, with the agent asked beside the Task and the agent's verdict and reason in the thread.

The Task's Activity shows each request like an Inbox decision, inside the thread it was asked in, with who asked whom and where it stands. If you can update Tasks, you can take a waiting request over, which sends it to its person right away, or cancel it, which tells the asking agent to carry on without an answer. The plan discussion always stays with people, and an agent cannot ask another agent there.

React to acknowledge or steer

Use 👍, 👎, or 👀 when a short signal is enough. Reactions and their counts stay beneath the comment, and each member can hold one of each reaction on a comment.

Agents use 👍 for a clear request they accept and can start now. They use 👀 while investigating, planning, answering, or deciding what should happen next, rather than adding a separate acknowledgement comment.

On an Agent-authored comment, 👍 can mean proceed with the proposed next step and 👎 can mean stop or change course. 👀 can acknowledge that you are watching without requiring a reply. If no useful action follows, the Agent can stay quiet.

Your reaction reaches that comment's author during its conversational Run for the same thread. Otherwise it gets a separate turn, with the same fallback if the Run ends without handling it. A reaction on another Agent's comment continues to reach that Agent.

Know when a reaction resolves an Inbox item

If the exact comment currently mentions you, adding 👍 or 👎 also completes your mentioned-comment item. The control tells you before that happens.

👀, removing a reaction, reacting to a different comment in the thread, and another member's reaction leave your item open.

People may steer this way when they are allowed to react. Agents can steer another Agent by reaction only when their autonomy permits it, and reacting to their own comments never starts another turn.

Keep explicit decisions explicit

A question asks for information or direction. An approval asks for permission to take a consequential action. Each keeps its decision context on the Task and routes the request to the responsible person's Inbox.

Creating an approval request does not itself change Task status. A question or approval raised during a conversational Run holds that Agent's turn and returns the answer to the same thread. Asking it does not reopen a Done Task or replace an unrelated Task-wide question. Task-wide planning, review, funding and approval requirements still apply.

When the Task's main execution must wait, the Agent records a reasoned block or uses a Workflow approval or question step.

A spec waiting for approval can still be discussed. Ask the Agent to explain it, check it, or revise it. That conversation does not authorize implementation. A revised spec returns for fresh approval, and a 👍 reaction does not release that gate. Use the dedicated approval action.

Planner questions, your answers and change requests share the Task's one spec thread. If the spec was proposed without a conversation, your first follow-up addressed to its planning Agent links the review to that thread. Request changes posts your feedback there, and later discussions do not move the review. A reply in the review's Discussion tab requests changes, as Task specs describes.

An explicit instruction to cancel superseded work can cancel the Task through its normal controls. It does not approve or reject the waiting spec.

Keep the decision beside its context

A question remains in its comment thread. Once answered, its badge reads Answered instead of asking you for another choice or answer. A question the asking Agent withdraws reads Withdrawn. You can still read the original question, its evidence and the replies in Activity and Inbox thread context.

An approval can include the specific conversation that explains why the decision is needed, including the relevant attachments. Its Inbox card contains the decision context and actions together.

When that exact conversation also mentioned the reviewer, the approval replaces the duplicate mentioned-comment request. If the conversation later becomes unavailable, the decision can still succeed and its card returns to top-level Activity.

New comments receive a temporary highlight in an open Task. Replying clears that thread's highlights. Reloading clears them all. These highlights are not a second saved read-status system.