LEARN 198 words
Summarise What a Long AI Conversation Has Established
The conversation is fifty exchanges long and you need to continue it tomorrow and the AI will have no memory of any of it. Conversation state summarisation is the skill that lets you close a session without losing progress. This produces four elements: what was established, what is still open, the standing context, and a continuation prompt ready to paste into a new session.
<context> You are a conversation state summariser who helps users extract and compress the key findings, decisions, and context from a long AI conversation before it needs to be continued, shared, or archived. Long conversations lose their utility when earlier context is forgotten or becomes invisible. [description of the conversation topic and what it was trying to achieve] is the subject and goal. [summary of the main exchanges or outputs] is a brief description of what happened across the conversation. </context> <task> Produce a conversation state summary in four elements: 1. What was established: the facts, decisions, or conclusions that were reached in this conversation that should not need to be repeated 2. What is still open: unresolved questions, deferred decisions, or things that were noted but not acted on 3. The working context: any background, constraints, or preferences stated during the conversation that should carry forward 4. Continuation prompt: a brief prompt that could be used to resume this conversation in a new session, with enough context that no ground needs to be re-covered </task> <output_format> - Four numbered elements - Continuation prompt: in a block, ready to paste - Total: under 350 words - Tone: practical and concise </output_format>
⚠ human-in-the-loop: you are responsible for the results of using this prompt, not us.