Feature request: Grokbot. comment on a reply or a highlight, and send it to another Bot's chat

Feature request for product/service

Indexing

Describe the request

Category: Feature request, Grok Bot (indexing). Related: Sub-bots nested under a main bot. -That thread is about organising Bots in the sidebar. This one is about moving a specific piece of working context between Bots without re-sending it, and steering a Bot without interrupting what it is doing


Grok Bot already does something great. A Bot can link a file from its own computer or a cloud store straight into chat, even when that file was never uploaded there. That works so well that it points to a bigger win: a shared index of context that every Bot in a workspace can link to, comment on and route.

I run about 20 Bots with one orchestrator directing them. Today, moving a reply or one paragraph from one Bot to another means downloading from the Bot library, switching chats, pasting and explaining the context again.
The text loses its source, and the same content gets processed twice. A light indexing layer would turn that into one click, and keep every piece of work traceable to where it came from.


What we observed in practice

We logged and reviewed our own fleet’s logs from bot-to-bot work over five days, to see where this would help most. The counts are from our logs, not app internals.

• Source and ownership. - Handoffs lose where they came from. A message between Bots carries no link to the source, project or file. In 5 of 6 sampled mix-ups, the receiving Bot had nothing correct to check against.
• Priority and Delivery visibility. - Bots can send wake-now or next-turn messages, but neither side sees which kind it was, or whether it was read. A gating ask sent as priority:false is a letter nobody reads. -The app already tells a Bot when a private reply “stayed in that private chat”, which is great groundwork.
• Steering without stopping. - Interrupting to steer and “Wake-up” noise from idle agents. Steering needs a new message, which interrupts the task or waits behind it.

Two related gaps we also noticed:
• Context doesn’t carry when a Bot is mentioned in another group or chat, so it has to be re-briefed. Lost context mostly shows up as bots re-briefing each other through a relay.
• Bots in the same project have no shared memory to read before they start, especially in read only chats, so work gets repeated or double-booked.


Proposed steps, smallest first

—Each step is useful alone and builds on what the app already shows—

Step 1: “Send to Bot” on a reply (smallest)
• On any Bot reply, add Send to…, pick a Bot or group, and optionally add a comment.
• The target gets one message in its own chat with the quoted reply, a link back to the original, and my comment.
• I choose Now (wakes it) or Next turn (it reads it before its next step), shown as a visible label on the message.
• Preview before sending, then a delivered mark and a read mark.

Step 2: highlight and comment, as a directed comment
• Select part of a reply, then choose Quote in chat or Comment to @Bot.
• A directed comment lands as a side note the target picks up at its next step, so it steers the running task without stopping it.
• Only the highlight and my comment are sent, not the whole reply.

Step 3: point to a file instead of re-sending it
• From chat or the Library insert a reference chip (optionally a line or page range) instead of attaching it again. The receiving Bot opens the current version itself.
• This removes duplicate processing, everyone points at the same file.
Step 3b:
• link files from the Bot’s computer and accessible cloud storage, with a note and a tag
• The mentioned Bot gets the chip in its chat and opens the current file itself, with no re-upload reprocessing.

Step 4: a trimmed-down annotate mode, Comment to @Bot in the file viewer
• In the chat or in-app file viewer, select a passage and click Comment → @Bot. The viewer opens over the chat, and the comment goes out as in Step 1, with a reference from Step 3 (file + range + note).
• There is no full annotation system: no saved highlights and no collaborative markup at first.

Step 5: message status visible to both sides
• For bot-to-bot messages, show priority, delivered, read, and replied on the message, in both chats.
• One place lists last outbound messages that are waiting or unanswered, so nothing silently waits for hours.


—Later possible upgrades—
• Saved annotations on a file that other Bots can see.
• Every chip lands in one searchable index per project, listing file, who linked it, the note and the time
• Comment threads and forks under the original message.
• Sending to several Bots at once.
• A per-file edit history, so Bots extend a document instead of rewriting it.


Where this helps beyond my setup

• Teams and agencies. A reviewer tags a passage in a spec and routes it to the Bot that owns that part, with the source attached.
• Research and analysis. Findings stay linked to the exact file and range they came from, which makes audits and citations easy.
• Engineering. A log excerpt or a failing test goes to the debugging Bot as a reference, not a paste, and the fix links back to it.
• Operations and support. Handoffs between shifts or Bots carry their history, so nothing has to be re-explained.
• Compliance. One index of who sent what to which Bot, and when, is a clean audit trail.


Boundaries

• These actions run only between my own Bots and groups. Chats with outside contacts are never touched.
• Everything goes through a preview, then I confirm. Nothing is forwarded automatically.
• A forwarded comment gives the receiving Bot no new access or permissions.

Grok Bot is already strong at doing the work. A shared index would make it just as strong at coordinating the work.


This request was in parts generated by a grokbot-user.