Copy an Issue as Markdown

An Issue in the Issue Tracker is more than its title — it has a state and a status, triage fields, a description, comments from whoever worked it, and often one or more linked conversations whose transcripts are the whole reason the issue exists. Getting all of that into a coding agent’s prompt, or into a ticket somewhere else, used to mean copying pieces by hand and losing the transcript. Copy full summary does it in one action: it puts the whole issue on the clipboard as a single Markdown document.

It’s a read-only action. It doesn’t change the issue, add a comment, or notify anyone — it only reads what’s already there and formats it.

Where to find it

There are two places to trigger it, and they produce the same document:

• On the issue detail page, open the overflow menu (⋯, labelled More actions) and choose Copy full summary.
• From a conversation’s Issues panel, each linked issue’s row has its own copy icon — its tooltip reads Copy full summary — so you don’t have to leave the conversation to grab the issue’s full context.

Don’t confuse it with the plain copy icon beside the issue title on the detail page: that one is Copy the in-app link, and it copies only the URL.

Because it’s a read action, it shows up for anyone who can see the issue at all — it does not require the manage-level permission that link and unlink actions need. If you have issues.view, you have this button; you don’t need issues.manage to use it, even where the panel is hiding management actions like unlink from you. (The Issue Tracker module also has to be switched on for your workspace at all — that’s a separate check from the permission.)

While the document is being assembled, the control shows a pending state and is disabled so you can’t click it twice: in the overflow menu the item’s text changes to Copying…, and the panel’s icon button simply greys out. Once it’s done, the control ticks — the menu item reads Copied, the panel’s icon becomes a green check — and reverts on its own after about a second and a half. The menu deliberately stays open so you can read the tick; there is no success toast.

If something goes wrong you get a toast rather than silence: Could not build the summary if the issue couldn’t be read, or Could not copy if the browser blocked clipboard access.

What the document contains

The Markdown is built from a fixed list of eight sections. Your workspace chooses which of them are copied and in what order under Settings → Issue Tracker → Copy Issue Summary (drag to reorder, clear a checkbox to leave a section out). A workspace that has never opened that page gets all eight, in the order below. Changing the selection needs issues.manage; with issues.view you can read the settings but not change them.

• Header — the top of the document, # Issue #<number> — <title>, followed by a field list: Type, State (open or closed), Status, Priority, Assignee, Reporter, Created, Updated, Environment, App version, Page, Page address, and Link. Type, Status and Priority are printed as your workspace’s own words for them, not as internal keys, so the document reads correctly to someone who has never seen your tracker. State is the one value that always tells you whether the issue is still open, whatever your workspace named its statuses.
• Triage — Module, Built, Scope, Reach, Customer priority, Flags (or none), and Duplicate of. Built, Scope and Reach are printed as the word plus what it means, because the reader of the pasted document has no legend.
• Description — the issue description as it was written.
• Labels — the free-form labels on the issue.
• Comments — the team’s notes on the issue, each with its author and timestamp. A comment is internal and team-only unless it was marked public, and the section says so in its own preamble: treat an internal comment as internal before the document goes anywhere a customer can see it. Public ones are marked inline.
• Linked conversations — for each conversation linked to the issue, a header block (Channel, Status, Contact, Started, Link) and then its transcript, not just a link to it. Each message is marked Inbound or Outbound; an internal note inside a conversation appears with a (note) suffix on the sender’s name rather than being hidden.
• Attachments — the files and screenshots on the issue, as absolute URLs.
• External reference — the external reference link, if the issue has one.

There is no separate provenance block. The details that say where the report came from — Environment, App version, Page (the page title), Page address (its URL) and Link (back to the issue) — are fields inside the Header section, so leaving Header out leaves those out too.

Any link in the document is built from the request that generated it, so it always points at the deployment you were actually using — not at some other environment’s configured base URL.

Empty is not the same as absent

An enabled section is always rendered, even when it holds nothing: it prints its heading with _None._ underneath. A section your workspace switched off is left out entirely. That distinction is deliberate — a reader who can’t tell “there are no labels” from “labels aren’t copied here” has to open the app to find out. Individual fields work the same way: a field with no value prints as — rather than disappearing.

When the document says it’s incomplete

A very long issue can’t always be copied in full, and the document says so rather than silently dropping content:

• Each transcript stops at 200 messages. If a conversation was cut off, a line directly under that transcript says only the first 200 messages are here and tells you to open the conversation for the rest. Individual message bodies are also cut at 4,000 characters.
• Only the first 20 linked conversations get a transcript. Past that, the section’s opening line states the total, how many follow in full, and how many are not in the document.
• The link list itself stops at 500. When an issue is at that bound, the section adds that the list stops at 500 and the issue may hold more — so the total above it is a floor, not a count.
• Comments are read under the same 500 bound. When they hit it, the section says the issue holds at least 500 notes and only the most recent 500 are here.
• Deleted links are called out, but only for the links the document actually attempted to read. If more than 20 conversations are linked and every one of the 20 it read has since been deleted, the wording says exactly that — it never claims to speak for the links past the cap.

If you paste the document somewhere and something reads incomplete, check the relevant section’s opening line or the end of that transcript first — the cap notice is there to tell you before you notice the hard way.