Skip to main content
Glama
100,867 servers. Updated

Matching MCP tools:

Matching MCP Connectors:

"Apple TV" matching MCP servers:

GET /v1/servers – MCP directory API reference
  • A
    license
    Not graded
    quality
    A
    maintenance
    Provides MCP clients typed tools for Apple Ads Platform research, reporting, auditing, and receipt-gated campaign management with a local-first, read-only-by-default safety model.
    5
    MIT
  • A
    license
    A
    quality
    A
    maintenance
    Owner-side Apple Podcasts analytics over the Podcasts Connect Reporter protocol: plays, unique and engaged listeners, followers, and per-episode listening for a show you own. Not the public iTunes catalog — this is the data Apple's hosting API does not expose.
    4
    26 npm
    1
    Apache 2.0
  • F
    license
    A
    quality
    B
    maintenance
    MCP server for historical asset analysis using tvDatafeed and cached CSV files. Enables fetching OHLCV bars, candlestick charts, and trade execution data for TradingView-style assets.
    3
    -
  • A
    license
    A
    quality
    A
    maintenance
    Enables podcast discovery and analysis by searching Apple's catalogue, retrieving chart rankings by country, reading listener reviews, parsing and validating RSS feeds, and pulling owner analytics.
    32
    38 npm
    MIT
  • A
    license
    Not graded
    quality
    C
    maintenance
    Cryptocurrency fundamental analysis tools via MCP. Enables users and AI agents to evaluate cryptocurrencies across 8 metrics using Sound Value principles. Provides educational material, fair value estimates, and valuation categories.
    1
    MIT
  • A
    license
    Not graded
    quality
    D
    maintenance
    Local-first MCP server that gives Claude Code web search, page reading, video transcription, and image analysis — without paid API keys. Runs SearXNG + whisper.cpp natively on Apple Silicon for zero-cost, low-latency research workflows.
    1
    MIT
  • A
    license
    Not graded
    quality
    Not graded
    maintenance
    An app intelligence query engine that enables analysis of over 1 billion reviews from Google Play and the Apple App Store. It provides tools for sentiment analysis, keyword rankings, competitive comparisons, and time-series forecasting across 250,000+ apps.
    16 npm
    2
    -
  • F
    license
    Not graded
    quality
    C
    maintenance
    Enables an MCP client to operate a personal Telegram account directly: read chats, send messages, interact with bots, schedule broadcasts, and research YouTube, Reddit, and Instagram.
    -
  • F
    license
    Not graded
    quality
    B
    maintenance
    Provides MCP tools to search and retrieve Apple App Store and Google Play app data, including app details, reviews, charts, developer info, and ASO keyword/rank estimates.
    -
  • F
    license
    Not graded
    quality
    D
    maintenance
    MCP server exposing Indonesia Stock Exchange (IDX) market data as tools — fundamentals, broker flow, company profiles, and technical analysis via TA-Lib.
    2
    -
  • A
    license
    A
    quality
    C
    maintenance
    An open-source MCP server for live Apple App Store competitor research, enabling AI agents to search apps, fetch metadata, compare competitors, and retrieve reviews and top charts as structured JSON.
    7
    16 PyPI
    MIT
  • A
    license
    A
    quality
    C
    maintenance
    Enables analyzing why a stock moved over a past window by running independent technical, company-news, and macro reads and reconciling them into verdicts and a narrative.
    1
    MIT
  • A
    license
    A
    quality
    B
    maintenance
    Gives AI agents live access to Apple App Store reviews: search the App Store, pull customer reviews for any iOS app, and turn them into insights like rating distribution, average rating per version, top complaint keywords, representative quotes, and side-by-side competitor comparisons.
    4
    MIT
  • A
    license
    A
    quality
    B
    maintenance
    Enables AI agents and humans to inspect a language model's internals through activations, sparse-autoencoder features, steering, and ablation, currently targeting Gemma 2 2B with Gemma Scope on Apple Silicon. The current server exposes run_prompt to report top SAE features per token and across text with Neuronpedia labels, with more tools planned.
    1
    MIT