mcpField
G
glim.sh - live data for any agent (hosted, Streamable HTTP)

glim.sh - live data for any agent (hosted, Streamable HTTP)

Remote
glim.shData & Analytics4Updated 9w ago

About glim.sh - live data for any agent (hosted, Streamable HTTP)

glim.sh is a hosted MCP server for agents that need live data from social platforms, websites, code repositories, commerce listings, video and public Telegram. Its tools also include AI-text detection.

Key features

  1. 01Search Twitter/X and Reddit
  2. 02Search and fetch web pages
  3. 03Search GitHub repositories and code
  4. 04Search Amazon listings and retrieve products
  5. 05Retrieve YouTube transcripts and Telegram data
  6. 06Score text for AI authorship

Use cases

  1. 01Research current social posts and web sources
  2. 02Inspect repositories, issues, or code
  3. 03Find Amazon products and pricing details
  4. 04Retrieve video transcripts or assess text authorship

MCP support

Provides a hosted Streamable HTTP MCP endpoint at https://glim.sh/mcp, exposing live-data and AI-text detection tools. MCP clients can restrict available tools using the endpoint’s tools parameter.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://glim.sh/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

14
ParameterTypeRequiredDescription
sortenumNoSort by relevance or recency: "top" (most engaged tweets matching the query - best for "what's the conversation about X") or "latest" (newest first - best for monitoring/recency). Default: "top".latest | top
querystringYesSearch query with operators (e.g. 'from:elonmusk AI min_faves:100 within_time:7d')
cursorstringNoPagination cursor from previous search. Ranking is strongest on page 1 - paged results under sort:'top' trend toward recency (upstream behavior).
formatenumNoOutput format. 'text' (default): compact human-readable view, fewer tokens. 'json': full structured data for machine parsing.text | json
end_datestringNoOnly tweets before this time. YYYY-MM-DD (inclusive through end of day UTC) or ISO 8601 datetime with Z/offset (e.g. 2026-04-13T14:30:00Z)
start_datestringNoOnly tweets on or after this time. YYYY-MM-DD (UTC midnight) or ISO 8601 datetime with Z/offset (e.g. 2026-04-13T14:30:00Z)
expand_urlsbooleanNoWhen true, auto-crawl entity URLs and attach crawled_content to tweets. Off by default: responses can grow by up to 4KB per expanded URL. Use glim_web_fetch(url) for targeted crawls instead.

Details

From the project README

glim.sh

Live data for any agent.

One remote MCP endpoint gives your agent live data from Twitter/X, Reddit, the open web, GitHub, Amazon, YouTube, and public Telegram channels, plus AI-text detection. No API keys, no scraping stack - connect the URL and pay per call.

https://glim.sh/mcp

Quick Start

# claude.ai - one click
open "https://claude.ai/customize/connectors?modal=add-custom-connector&connectorName=glim&connectorUrl=https%3A%2F%2Fglim.sh%2Fmcp"

# Claude Code
claude mcp add -s user -t http glim https://glim.sh/mcp

# Claude Code plugin (MCP server + usage skill, auto-updates)
claude plugin marketplace add glim-sh/glim-mcp
claude plugin install glim@glim

# Codex
codex mcp add glim --url https://glim.sh/mcp

# OpenCode (interactive: choose remote, paste the URL)
opencode mcp add

# Agent skill (any skills-aware agent)
npx skills add https://glim.sh --yes

# No account? Pay per call from a local crypto wallet (x402)
npx x402-proxy mcp add glim https://glim.sh/mcp

Any other MCP client: add the endpoint to your config.

{
  "mcpServers": {
    "glim": {
      "url": "https://glim.sh/mcp"
    }
  }
}

On first use the client runs a browser OAuth sign-in. Manage your balance at glim.sh/app. The wallet path needs no account at all: calls auto-pay USDC (Base or Solana) on x402 challenges; MPP on Tempo is accepted on the same endpoint.

Tools

ToolPriceUse for
glim_twitter_search$0.005Search tweets with advanced operators
glim_twitter_get$0.005Tweet with thread context, or user profile, by URL/id/handle
glim_reddit_search$0.01Search Reddit posts
glim_reddit_get$0.015Post + comments, subreddit feed, or user activity
glim_web_search$0.01Semantic web search
glim_web_fetch$0.002Clean page extraction (SSR, SPA shells, PDFs)
glim_github_search$0.002Search GitHub repos, conversations, or code
glim_github_get$0.002Repo metadata, files, history, PRs, issues
glim_amazon_search$0.005Search Amazon listings with prices, ratings, ASINs
glim_amazon_get$0.01Amazon product detail by ASIN
glim_youtube_get$0.01YouTube subtitles/transcript by video id
glim_telegram_search$0.01Discover public Telegram channels by name or topic
glim_telegram_get$0.005Channel posts, in-channel history search, or post + discussion comments
glim_detect_aifrom $0.06Score text for AI authorship, with a per-segment breakdown

Restrict the tool list per session with ?tools=:

https://glim.sh/mcp?tools=glim_twitter_search,glim_web_fetch

REST API

Every tool is also a plain REST endpoint under https://glim.sh/api/v1/, gated by the same payment challenges. Full spec: glim.sh/openapi.json.

npx x402-proxy https://glim.sh/api/v1/twitter/get/cascade_fyi

Documentation

For full documentation, visit glim.sh/docs.

About this repo

This repo holds the registry manifests, the Claude Code plugin, and documentation for the glim MCP server - not the server source. Open an issue for listing or documentation fixes, or email [email protected].

License

MIT (c) 2026 Misha Kolesnik

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.