Uh oh!
There was an error while loading. Please reload this page.
Generate system instructions from chat history - #403
Conversation
🦋 Changeset detectedLatest commit: 243929c The changes in this PR will be included in the next version bump. This PR includes changesets to release 3 packages
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
b3f5f6a to
5e08c13CompareUh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
c7fe4cd to
9f22d41CompareUh oh!
There was an error while loading. Please reload this page.
Users can promote a working conversation into an editable draft instead of writing the agent prompt from scratch, and nothing is saved until they apply or copy it. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Walk a turn's user messages newest-first so the transcript reverse keeps them in order.
9f22d41 to
0672b2fCompareUh oh!
There was an error while loading. Please reload this page.
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 74e33f4. Configure here.
Uh oh!
There was an error while loading. Please reload this page.
Co-authored-by: Cursor <cursoragent@cursor.com>

Fixes#357
System instructions currently have to be written from scratch, even when the chat already made it obvious how the user wants the agent to behave.
This adds a From chat button on the session header. It drafts instructions from the transcript, you can edit them in a dialog, and nothing is saved until you apply. Inline chats can apply the draft to that session. Named/registry agents only get copy, so we don't mutate the saved agent.
Short chats return 422 before we call the model, so a "hey"/"hi" conversation doesn't turn into a generic helpful-assistant prompt.
New route:
POST /api/v1/sessions/{session_id}/generate-instructionsIt only returns a suggestion (
instructions,current_instructions,sources). Saving still goes through the existing session update.I didn't regenerate the SDK. The UI adapter hits the new path with
client.fetch.Note
Medium Risk
New authenticated session endpoint invokes the session model on transcript data and updates agent specs on apply; mistakes could affect future turns, though drafts are explicit and short chats are gated.
Overview
Adds From chat in the session header so users can infer system instructions from the transcript, edit them in a modal, and only persist when they choose Apply (inline/mutable sessions). Named registry-bound chats get Copy only so saved agents are not mutated server-side.
Backend exposes
POST /api/v1/sessions/{session_id}/generate-instructions, which returns a draft (instructions,current_instructions,sources) without writing the session. Transcript extraction skips subagent threads; 422 is returned when user text is too thin before any LLM call, and model output is filtered to block generic one-liners. The harness UI adapter calls the new route viaclient.fetch; saving still uses the existing session update path after flush/adopt.Reviewed by Cursor Bugbot for commit 243929c. Bugbot is set up for automated code reviews on this repo. Configure here.