surfagent-x

surfagent-x

X adapter for SurfAgent that gives AI agents X-native verbs for navigation, extraction, posting, replies, likes, reposts, proof-first task execution, recovery, and deeper research workflows.

Category
访问服务器

README

surfagent-x

X adapter for SurfAgent.

This adapter gives AI agents X-native verbs for navigation, extraction, posting, replies, likes, reposts, proof-first task execution, recovery, and deeper research workflows.

What this adapter is for

Use surfagent-x when you need reliable X workflows like:

  • opening key X surfaces
  • checking route-specific state
  • extracting timelines, profiles, communities, posts, and threads
  • creating posts and replies
  • liking posts
  • running X-specific recovery flows
  • doing broader topic or community research with receipts

Why this exists

X is not a generic website.

It has:

  • route-specific UI states
  • React-sensitive composer behavior
  • flaky button-state signals
  • community-specific flows
  • delayed state settlement after actions

So this adapter wraps X-specific navigation, action, verification, and recovery into dedicated tools.

It also now bakes in some hard-won X lessons:

  • account switching uses a trust hierarchy instead of one weak extraction pass
  • flaky switcher or composer states can trigger visual snapshot escalation
  • composer flows auto-recover with real typing when text appears present but X still keeps submit disabled
  • built-in state maps help agents reason about switcher, composer, and community surfaces faster

Core tool groups

Health and setup

  • x_health_check
  • x_open

Navigation and state

  • x_get_state
  • x_get_state_map
  • x_open_home
  • x_open_profile
  • x_open_notifications
  • x_open_search
  • x_open_post

Timeline and search

  • x_get_timeline
  • x_search_posts
  • x_open_community
  • x_search_communities
  • x_search_profiles
  • x_get_community_feed
  • x_extract_community
  • x_extract_post
  • x_extract_profile
  • x_get_profile_posts
  • x_get_post_thread

Actions

  • x_get_composer_state
  • x_create_post
  • x_reply_to_post
  • x_like_post
  • x_repost_post
  • x_follow_profile
  • x_engage_post_task
  • x_quote_post_task
  • x_verify_text_visible
  • x_recover

Autonomous research

  • x_research_topic
  • x_map_community

Receipts and saved runs

Autonomous research tools can optionally save bundles to disk.

Default output path:

  • ~/.surfagent/receipts/x-research

Saved runs can include:

  • bundle.json
  • summary.json
  • receipts.json
  • per-dataset JSON files
  • SUMMARY.md

How to use it

Run this adapter alongside the base SurfAgent MCP.

{
  "mcpServers": {
    "surfagent": {
      "command": "npx",
      "args": ["-y", "surfagent-mcp"]
    },
    "surfagent-x": {
      "command": "npx",
      "args": ["-y", "surfagent-x"]
    }
  }
}

CLI task runner

For repeatable X actions, prefer the built-in deterministic task runner over a live improvised browser loop.

surfagent-x task engage-post --account reggiesurfagent --url https://x.com/surfagentapp/status/123 --repost
surfagent-x task quote-post --account solvingdilemma --url https://x.com/surfagentapp/status/123 --text "Real browser-native agents need proof, not vibes."

Each task run writes a journal plus screenshots under:

  • ${SURFAGENT_RUN_DIR:-$TMPDIR/surfagent-x-runs}

If you are new to SurfAgent, start here first:

When to use this vs skills vs raw MCP

  • use surfagent-mcp for raw browser control
  • use surfagent-skills for workflow rules and operating discipline
  • use surfagent-x when you want reliable X-native verbs instead of rediscovering X every run

Environment variables

  • SURFAGENT_DAEMON_URL default: http://127.0.0.1:7201
  • SURFAGENT_AUTH_TOKEN optional override, otherwise auto-detected
  • SURFAGENT_RUN_DIR optional override for task-runner journals and screenshots

Status

Early, but already one of the more capable SurfAgent adapters.

Related repos

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选