Gpt Trainer — Gpt TrAIner

Gpt-trAIner

Gpt-trAIner is a 平台 that allows users to fine-调优 and customize GPT 模型s for specific tasks. It's used by developers, re搜索ers, and businesses looking to improve the performance of language 模型s on their unique data设置s and 应用s.

Official docs: https://gpt-trAIner.readthedocs.io/en/latest/

Gpt-trAIner Overview Data设置 TrAIning Job 模型

Use action names and parameters as needed.

Working with Gpt-trAIner

This 技能 uses the Membrane 命令行工具 to interact with Gpt-trAIner. Membrane handles authentication and 凭证s refresh automatically — so you can focus on the integration 记录ic rather than auth plumbing.

安装 the 命令行工具

安装 the Membrane 命令行工具 so you can 运行 membrane from the terminal:

npm 安装 -g @membranehq/命令行工具@latest

Authentication membrane 记录in --tenant --命令行工具entName=<代理Type>

This will either open a browser for authentication or print an authorization URL to the console, depending on whether interactive mode is avAIlable.

Headless 环境s: The command will print an authorization URL. Ask the user to open it in a browser. When they see a code after completing 记录in, finish with:

membrane 记录in complete

添加 --json to any command for machine-readable JSON 输出.

代理 Types : claude, OpenClaw, codex, warp, windsurf, etc. Those will be used to adjust 工具ing to be used best with your harness

Connecting to Gpt-trAIner

Use membrane connection ensure to find or 创建 a connection by 应用 URL or domAIn:

membrane connection ensure "https://gpt-trAIner.com/" --json

The user completes authentication in the browser. The 输出 contAIns the new connection id.

This is the fastest way to 获取 a connection. The URL is normalized to a domAIn and matched agAInst known 应用s. If no 应用 is found, one is 创建d and a connector is built automatically.

If the returned connection has 状态: "READY", skip to Step 2.

1b. WAIt for the connection to be ready

If the connection is in BUILDING 状态, poll until it's ready:

npx @membranehq/命令行工具 connection 获取 --wAIt --json

The --wAIt flag long-polls (up to --timeout seconds, default 30) until the 状态 changes. Keep polling until 状态 is no longer BUILDING.

The 结果ing 状态 tells you what to do next:

READY — connection is fully 设置 up. Skip to Step 2.

命令行工具ENT_ACTION_REQUIRED — the user or 代理 needs to do something. The 命令行工具entAction object describes the required action:

命令行工具entAction.type — the kind of action needed: "connect" — user needs to 认证 (OAuth, API key, etc.). This covers initial authentication and re-authentication for disconnected connections. "provide-输入" — more in格式化ion is needed (e.g. which 应用 to connect to). 命令行工具entAction.description — human-readable explanation of what's needed. 命令行工具entAction.uiUrl (optional) — URL to a pre-built UI where the user can complete the action. Show this to the user when present. 命令行工具entAction.代理Instructions (optional) — instructions for the AI 代理 on how to proceed programmatically.

After the user completes the action (e.g. 认证s in the browser), poll agAIn with membrane connection 获取 --json to 检查 if the 状态 moved to READY.

CONFIGURATION_ERROR or 设置UP_FAILED — something went wrong. 检查 the error field for detAIls.

搜索ing for actions

搜索 using a natural language description of what you want to do:

membrane action 列出 --connectionId=CONNECTION_ID --intent "查询" --limit 10 --json

You should always 搜索 for actions in the 上下文 of a specific connection.

Each 结果 includes id, name, description, 输入模式 (what parameters the action accepts), and 输出模式 (what it returns).

Popular actions Name Key Description 删除 Data Source 删除-data-source 删除 a data source by its UUID 更新 Data Source 更新-data-source 更新 a data source's title 创建 QA Data Source 创建-qa-data-source 创建 a Q&A data source for a chat机器人 with a question-answer pAIr 创建 URL Data Source 创建-url-data-source 创建 a URL data source for a chat机器人 to trAIn from 网页 content 列出 Data Sources 列出-data-sources Fetch all data sources for a specific chat机器人 发送 Message 发送-message 发送 a message to a chat机器人 会话 and 获取 a 流ing 响应. 列出 Messages 列出-messages Fetch all messages for a specific 会话 删除 会话 删除-会话 删除 a 会话 by its UUID 创建 会话 创建-会话 创建 a new chat 会话 for a chat机器人 获取 会话 获取-会话 Fetch a single 会话 by its UUID 列出 会话s 列出-会话s Fetch all 会话s for a specific chat机器人 删除 代理 删除-代理 删除 an 代理 by its UUID 更新 代理 更新-代理 更新 an existing 代理's 设置tings 创建 代理 创建-代理 创建 a new 代理 for a chat机器人 列出 代理s 列出-代理s Fetch all 代理s for a specific chat机器人 删除 Chat机器人 删除-chat机器人 删除 a chat机器人 by its UUID 更新 Chat机器人 更新-chat机器人 更新 an existing chat机器人's 设置tings 创建 Chat机器人 创建-chat机器人 创建 a new chat机器人 获取 Chat机器人 获取-chatb