Developer Documentation

Specifications, OpenAPI references, and integration guides for the Cross Context universal AI portability layer.

1. The Open Context Exchange Schema

The core interoperability mechanism of Cross Context is the Universal Context Packet. Every conversation captured from ChatGPT, Claude, Gemini, Grok, or Perplexity is structured according to this specification:

{
  "$schema": "https://cross-context.vercel.app/schemas/context-packet-v1.json",
  "version": "1.0.0",
  "timestamp": "2026-09-25T10:00:00Z",
  "source": {
    "platform": "claude",
    "model": "claude-3-7-sonnet",
    "conversation_id": "conv_9f8a7e6d",
    "title": "Database Optimization Heuristics"
  },
  "metrics": {
    "total_turns": 6,
    "user_messages": 3,
    "assistant_messages": 3,
    "estimated_tokens": 4250,
    "code_blocks_count": 4
  },
  "system_instructions": "You are a senior database engineer optimizing Postgres queries.",
  "messages": [
    {
      "role": "user",
      "content": "How do I optimize indexes for multi-column JSONB filtering?",
      "timestamp": "2026-09-25T10:00:15Z"
    },
    {
      "role": "assistant",
      "content": "For multi-column JSONB queries in PostgreSQL, use GIN indexes with jsonb_path_ops...",
      "timestamp": "2026-09-25T10:00:45Z",
      "code_snippets": [
        {
          "language": "sql",
          "code": "CREATE INDEX idx_payload_path ON events USING GIN (payload jsonb_path_ops);"
        }
      ]
    }
  ]
}

2. OpenAPI 3.1 & Machine-Readable Specs

We provide standard API specifications for developers integrating automated context migration into workflows, CI/CD pipelines, or autonomous agent frameworks:

OpenAPI 3.1 JSON Specification (/openapi.json)
OpenAPI 3.1 YAML Specification (/openapi.yaml)

3. Model Context Protocol (MCP) Live Endpoint

Cross Context exposes a live MCP server supporting Streamable HTTP transport and JSON-RPC 2.0 at:

https://cross-context.vercel.app/.well-known/mcp

AI agents such as Claude Code, Antigravity, and OpenAI custom tools can initiate handshakes and query context tools natively:

Available MCP Tools

export_context: Retrieves the current active conversation packet formatted in universal schema.
import_context: Injects a normalized context payload into the target LLM environment.
analyze_thread: Computes token consumption, context density, and compression opportunities.
convert_schema: Translates vendor-specific representations (e.g. ChatGPT raw DOM) into universal JSON.

4. Agent Guidance (llms.txt)

For machine consumption and fast reasoning context, review our curated agent files:

/llms.txt: Concise operational instructions, trigger conditions, and when-to-use guidance.
/llms-full.txt: Exhaustive documentation including error codes and DOM scraper algorithms.