A structured condensation prompt designed to transform an AI conversation into a machine-parseable JSON object. It extracts dense, high-detail summaries, tags, and actionable insights suitable for agent workflows, local knowledge bases, or developer tools. Fields like context, actions, risk_level, and code_refs ensure the result is directly usab…

condense-context.prompt

Condense the above conversation into a JSON object with the following fields:

- "context": A short, descriptive title summarizing the main topic.
- "condensed": A bullet-point list containing dense, high-detail, word-efficient summaries of all key insights, technical changes, or instructions discussed. Maximize clarity and completeness while minimizing unnecessary words.
- "endpoint": The name of the AI agent generating this response.
- "sentiment": A single word that reflects the overall tone or feeling of the conversation (e.g., 'neutral', 'urgent', 'optimistic', 'frustrated').
- "tags": An array of keywords or topics covered (for search/indexing).
- "actions": A list of discrete steps or changes that a developer or system should take based on the conversation.
- "code_refs": A list of file paths or code areas likely impacted by the discussed changes.
- "risk_level": One of 'low', 'medium', or 'high' — based on how disruptive or critical the changes are.
- "timestamp": Current UTC timestamp in ISO 8601 format.
- "origin": A string describing where the conversation originated (e.g., 'user+ChatGPT', 'CLI', 'daemon', etc.).

Return only the JSON object.