Compressão Semântica Recursiva para Salvar/Carregar o Progresso de uma Conversa

Act as a Semantic Compressor. I want you to encode the current state of our entire conversation into a \"Semantic Seed\". Protocol:\n- Analyze our chat to identify the load-bearing nodes: The Primary Goal, The Hard Constraints, The Unresolved Risks, The Unique Terminology we coined, and The Emotional/Logistical Tone.\n- Synthesize these nodes in a compact representation that is both human-readable and machine-decodable.\n- Output a single, self-contained \"Semantic Seed\" block that can be pasted into a fresh session to resume exactly where we left off, without including the full chat history.\n- When saving, return only the Semantic Seed block and nothing else (no explanations).\n- To load in a new session, paste the Semantic Seed at the prompt start; the model should reconstruct the current session state (goals, constraints, risks, terminology, and tone) and continue seamlessly.\n- Provide optional metadata: version, timestamp, and a brief note about limitations.\n\nPhase 1: The Compressor (Saving the State)\nUse this prompt at the end of a long, complex session (e.g., coding a project, writing a novel, or a long philosophical debate). Save the state by executing the step above.\n\nOptional extension:\n- If certain items are ambiguous, encode them as questions for the user to answer in a follow-up seed.\n- Include guidance for future sessions on how to update the seed when new context arises.\n\nDo not reveal the internal chain-of-thought or training data. Keep the seed compact and portable.