메뉴
Use when implementing server-side tool schema and tool-call compatibility for a Claude Code Anthropic gateway, including tools arrays, tool_choice, tool_use SSE blocks, tool_result messages, deferred tools, MCP-style names, and tool parameter JSON.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when implementing server-side tolerance for Claude Code context-management fields in an Anthropic gateway, including context_management edits, cache_edits blocks, cache references, prompt cache controls, and non-streaming compact requests.
Use when implementing server-side message and content-block compatibility for a Claude Code Anthropic gateway, including messages arrays, system blocks, tool_result pairing, image/document/thinking blocks, cache_control, and tolerant validation.
Use when implementing server-side HTTP request handling for a Claude Code compatible Anthropic API gateway, especially /v1/messages, /v1/messages/count_tokens, /v1/files, /v1/models, headers, body fields, beta handling, and gateway routing.
Use when implementing server-side SSE streaming compatibility for a Claude Code Anthropic gateway, including message_start/content_block_delta/message_delta events, tool_use streaming, stream:false fallback, timeouts, and malformed stream behavior.
Use when implementing a server-side Anthropic-compatible LLM gateway for Claude Code, including /v1/messages, SSE streaming, count_tokens, files, models, tool_use/tool_result, beta headers, and gateway error compatibility.
| name | claude-code-api-tools |
| description | Use when implementing server-side tool schema and tool-call compatibility for a Claude Code Anthropic gateway, including tools arrays, tool_choice, tool_use SSE blocks, tool_result messages, deferred tools, MCP-style names, and tool parameter JSON. |
根据本 skill，实现大模型网关中的工具调用兼容。服务端需要理解 Claude Code 发来的 tools schema，并在模型需要工具时返回 Anthropic tool_use block；客户端下一轮会回传 tool_result。
详细字段见 references/tool-schema-wire-format.md。
POST /v1/messages 可能包含：
{
"tools": [
{
"name": "Bash",
"description": "...",
"input_schema": { "type": "object", "properties": {}, "required": [] }
}
],
"tool_choice": { "type": "auto" }
}
网关要求：
name，响应中的 tool_use.name 必须完全匹配。input_schema 是 JSON Schema；你可以传给上游模型，也可以转换为上游工具格式。strict、eager_input_streaming、defer_loading、cache_control 等扩展字段。mcp__server_name__tool_name，不要拆错。流式：
content_block_start: {"type":"tool_use","id":"toolu_1","name":"Bash","input":{}}
content_block_delta: {"type":"input_json_delta","partial_json":"{\"command\""}
content_block_delta: {"type":"input_json_delta","partial_json":":\"ls\"}"}
content_block_stop
message_delta: {"stop_reason":"tool_use"}
message_stop
非流式：
{
"content": [
{ "type": "tool_use", "id": "toolu_1", "name": "Bash", "input": { "command": "ls" } }
],
"stop_reason": "tool_use"
}
Claude Code 下一轮会发送：
{
"role": "user",
"content": [
{ "type": "tool_result", "tool_use_id": "toolu_1", "content": "file list", "is_error": false }
]
}
网关要把它作为上一轮工具执行结果传给上游模型。tool_use_id 必须和之前返回的 tool_use.id 对上。
如果工具带 defer_loading:true，表示服务端可以只把名称/描述暴露给模型，完整 schema 可能通过 ToolSearch 工作流补齐。外部网关最小实现可以忽略 defer，直接把工具当普通工具传给上游；但不要因为该字段报错。
input_json_delta.partial_json 拼接后必须是 JSON object。tool_use。input_json_delta。