synap
Synap (pool.linkrra.com/v1), Linkrra's OpenAI-compatible LLM API, as an MCP server.
我該用這個嗎
品質與安全性
B
根據工具定義與協定合規性的自動化分析。
上下文成本
~207Token(工具定義)
~666 B典型回應大小
極小的注意力影響(128k 上下文的 0.16%)
這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高,可用於其他工作的注意力就越少。
安裝
一鍵安裝
將以下內容加入你的 `claude_desktop_config.json` 檔案:
{
"mcpServers": {
"synap": {
"url": "https://pool.linkrra.com/mcp"
}
}
}遠端端點
https://pool.linkrra.com/mcpstreamable-http它能做什麼
工具清單
工具(2)
🟢 唯讀🟡 寫入🔴 刪除⚪ 未知
⚪synap_chat_completion(model, messages, temperature, max_tokens)
Call Synap, Linkrra's OpenAI-compatible inference API (pool.linkrra.com/v1), for a chat completion. Requires the caller to supply their own Synap API key as the Authorization: Bearer header on the MCP request.
輸入結構描述
{
"type": "object",
"properties": {
"model": {
"type": "string",
"description": "Model id — see synap_list_models for the catalogue."
},
"messages": {
"type": "array",
"items": {
"type": "object",
"properties": {
"role": {
"type": "string",
"enum": [
"system",
"user",
"assistant"
]
},
"content": {
"type": "string"
}
},
"required": [
"role",
"content"
]
}
},
"temperature": {
"type": "number"
},
"max_tokens": {
"type": "integer"
}
},
"required": [
"model",
"messages"
]
}🟢synap_list_models
List models currently available on Synap (pool.linkrra.com/v1/models).
輸入結構描述
{
"type": "object",
"properties": {}
}社群
證據
近期觀測
已驗證未記錄版本2 個工具
已驗證未記錄版本2 個工具