Conversation Example

A normalized conversation imported from Claude, demonstrating the PAM conversation schema. This corresponds to conv-001 in the Complete Memory Store example.

Example

{
  "schema": "portable-ai-memory-conversation",
  "schema_version": "1.0",
  "id": "conv-001",
  "provider": {
    "name": "claude",
    "conversation_id": "28d595a3-5db0-492d-a49a-af74f13de505",
    "account_id": "acc-uuid",
    "export_format_version": "2025.01"
  },
  "title": "Initial setup and infrastructure discussion",
  "temporal": {
    "created_at": "2024-06-01T10:00:00Z",
    "updated_at": "2024-06-01T16:00:00Z"
  },
  "participants": [
    {
      "role": "user",
      "name": "User"
    },
    {
      "role": "assistant",
      "name": "Claude"
    }
  ],
  "messages": [
    {
      "id": "msg-001",
      "provider_message_id": "uuid-from-claude",
      "role": "user",
      "content": {
        "type": "text",
        "text": "Preciso configurar a infraestrutura de rede do datacenter com isolamento VLAN e BGP."
      },
      "created_at": "2024-06-01T10:00:00Z",
      "parent_id": null,
      "children_ids": [
        "msg-002"
      ],
      "model": null,
      "is_thought": false,
      "token_count": 18,
      "attachments": [],
      "citations": [],
      "tool_calls": [],
      "raw_metadata": {}
    },
    {
      "id": "msg-002",
      "provider_message_id": "uuid-from-claude-2",
      "role": "assistant",
      "content": {
        "type": "text",
        "text": "Vamos definir a topologia de rede começando pelo isolamento de VLANs por tenant e a configuração BGP para upstream."
      },
      "created_at": "2024-06-01T10:01:00Z",
      "parent_id": "msg-001",
      "children_ids": [],
      "model": "claude-3-opus-20240229",
      "is_thought": false,
      "token_count": 52,
      "attachments": [],
      "citations": [],
      "tool_calls": [],
      "raw_metadata": {}
    }
  ],
  "model": "claude-3-opus-20240229",
  "system_instruction": null,
  "is_archived": false,
  "tags": [
    "infrastructure",
    "networking",
    "datacenter"
  ],
  "import_metadata": {
    "importer": "pam-converter/1.0.0",
    "importer_version": "claude-importer/2025.01",
    "imported_at": "2026-02-15T22:00:00Z",
    "source_file": "conversations.json",
    "source_checksum": "sha256:abcdef1234567890abcdef1234567890abcdef1234567890abcdef1234567890"
  }
}

Key Sections

Provider

Identifies the source platform (claude), the original conversation ID as assigned by Claude (28d595a3-…), and the export format version (2025.01). The export_format_version is used by importers to select the correct parsing logic, since provider export formats may change over time.

Participants

Two participants with normalized roles: user and assistant. PAM normalizes provider-specific role values (Claude exports use human/assistant; ChatGPT uses user/assistant) into a consistent vocabulary.

Messages

Two messages demonstrating the thread structure:

• msg-001 — User message with parent_id: null (root message) and children_ids: ["msg-002"] linking to the reply. The user message has model: null since it was not generated by a model.
• msg-002 — Assistant reply with parent_id: "msg-001" pointing back to the user message, model: "claude-3-opus-20240229" identifying the model that generated it, and empty children_ids indicating it is a leaf node.

Claude conversations are linear (no branching), so parent_id/children_ids form a simple chain. For branching conversations (like ChatGPT), a message may have multiple children, each representing a different conversation fork. The PAM conversation schema supports both linear and DAG structures using the same fields.

Import Metadata

Records the importer version (claude-importer/2025.01), source file (conversations.json), and a SHA-256 checksum of the source file for traceability. This allows re-import detection and auditing of which source export a conversation was derived from.