Model Context Protocol

ComparEdge MCP Servers

Technical reference for the ComparEdge MCP family: three zero-dependency servers that give AI assistants and IDEs verified SaaS pricing, live LLM token costs, and pricing-change tracking. Pick a server below.

ComparEdge MCP Server

Flagship@comparedge/mcp-server · v2.7.1

Verified pricing, alternatives, and side-by-side comparisons for 495 SaaS and AI tools across 44 categories. Ask in plain English, get structured data with links back to the source.

Tools

Prompts

Dependencies

terminal

npx @comparedge/mcp-server

Installation

Runs via npx with no global install. Node.js 18 or later required. Add the block for your client, then restart it.

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "comparedge": {
      "command": "npx",
      "args": ["-y", "@comparedge/mcp-server"]
    }
  }
}

Cursor

~/.cursor/mcp.json

{
  "mcpServers": {
    "comparedge": {
      "command": "npx",
      "args": ["-y", "@comparedge/mcp-server"]
    }
  }
}

VS Code (GitHub Copilot)

.vscode/mcp.json

{
  "servers": {
    "comparedge": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@comparedge/mcp-server"]
    }
  }
}

Configuration

No environment variables or API keys. All data is fetched from public ComparEdge endpoints at runtime and cached in memory.

Parameter	Value	Notes
command	npx	Node.js package runner, no global install needed
args[0]	-y	Auto-confirm flag, skips interactive prompts
args[1]	@comparedge/mcp-server	Package name on the NPM registry
type	stdio	Transport protocol; only stdio is supported

Tools Reference

The ComparEdge server exposes 8 MCP tools. Each returns structured plain text.

search_toolstool

Search 495 SaaS tools by name, category, keyword, or natural language. Stopwords are stripped, keywords matched across name, slug, category, and description.

Parameters

querystringrequiredProduct name, category, keyword, or natural language (e.g. "best CRM for startups")

limitnumberoptionalMaximum number of results (default: 5)

Returns

Ranked list with name, category, starting price, free tier indicator, and a profile link.

get_tooltool

Retrieve the complete profile for a specific tool by its slug.

Parameters

slugstringrequiredURL-safe product identifier (e.g. notion, hubspot, linear)

Returns

Description, category, pricing summary, rating, free plan status, and links to the pricing matrix and alternatives.

get_pricingtool

Retrieve verified pricing plans for a specific tool.

Parameters

slugstringrequiredURL-safe product identifier

Returns

Plan names, prices, billing intervals, tier highlights, free trial availability, and a link to the full pricing matrix.

get_alternativestool

Find verified alternatives to a tool within the same category.

Parameters

slugstringrequiredIdentifier of the reference product

limitnumberoptionalMaximum results (default: 5, max: 10)

Returns

Ranked alternatives with pricing, ratings, free tier status, and per-item comparison links.

compare_toolstool

Side-by-side structured comparison of two software products.

Parameters

tool1stringrequiredSlug of the first tool

tool2stringrequiredSlug of the second tool

Returns

Structured comparison of pricing, features, ratings, and key differences.

list_categorytool

Browse all tools in a category with a pricing overview.

Parameters

categorystringrequiredCategory slug (e.g. "llm", "ai-coding", "crm")

sort_bystringoptional"rating" (default) or "startingPrice"

free_onlybooleanoptionalOnly tools with a free plan (default: false)

Returns

Tools in the category with name, pricing, rating, and free plan status.

get_leaderboardtool

Top-rated tools by category, ranked by aggregated review scores.

Parameters

categorystringoptionalCategory slug, or "all" for the overall leaderboard (default: "all")

limitnumberoptionalNumber of top tools (default: 10)

Returns

Ranked tools with aggregated rating scores, pricing, and profile links.

list_categoriestool

List all 44 supported software categories with slugs and display names.

Returns

Category slugs and display names. Use slugs as input to list_category or get_leaderboard.

Prompts Reference

Built-in prompts are guided workflows that chain multiple tools automatically. Select a prompt in your AI client to start a structured session.

find_best_toolprompt

Guided search for the best tool matching a use case. Calls search_tools, then get_pricing on top results.

Arguments

use_casestringrequiredWhat you need to do (e.g. "manage customer relationships")

compare_pricingprompt

Side-by-side pricing comparison of two tools. Calls compare_tools plus get_pricing on both.

Arguments

tool_astringrequiredFirst tool name or slug

tool_bstringrequiredSecond tool name or slug

evaluate_toolprompt

Full evaluation in one session: get_tool for the profile, get_pricing for plans, get_alternatives for competitors.

Arguments

toolstringrequiredTool name or slug

category_overviewprompt

Overview of the best tools in a category. Resolves the slug, calls get_leaderboard, then get_pricing on the top 3.

Arguments

categorystringrequiredSoftware category (e.g. "CRM", "project management")

Usage Examples

Natural language queries supported by any MCP-compatible AI assistant after the server is configured.

Pricing queries

›“What are the pricing plans for Linear?”

›“Does HubSpot offer a free plan?”

›“What is the cheapest CRM under $20/month?”

›“Compare Figma vs Sketch pricing”

Alternatives

›“Find alternatives to Slack for a team of 10”

›“What are the best open-source alternatives to Jira?”

›“Show me tools similar to Linear”

›“Suggest a cheaper replacement for Salesforce”

Discovery

›“List all project management tools with a free tier”

›“What AI writing tools are available under $30/month?”

›“Find DevOps tools that support self-hosting”

GitHub Copilot

The ComparEdge MCP server integrates with GitHub Copilot Chat via VS Code MCP extension support.

Copilot Chat integration

Once configured, query ComparEdge directly from the Copilot Chat panel in VS Code:

Copilot Chat

@workspace What are the pricing plans for Linear?
@workspace Find alternatives to Notion for documentation
@workspace Compare GitHub Actions vs CircleCI pricing

Copilot Extension endpoint

A native Copilot Extension endpoint is available for GitHub App integrations:

POSThttps://comparedge.com/api/copilotextension messages

GEThttps://comparedge.com/api/copilothealth check

Architecture

Single Node.js module communicating over stdio. No runtime dependencies beyond Node.js built-in modules.

Transportstdio (MCP standard)

RuntimeNode.js 18+

DependenciesNone (built-ins only)

Data sourcescomparedge.com/llms-*.json

Cache strategyIn-memory, TTL 6h

MCP specification2025-03-26

Changelog

Version history for the @comparedge/mcp-server package.

v2.7.12026-07-11

-Version bump aligned with the MCP family release (ComparEdge, LLM Cost, Price Watch published together)
-Docs restructured into a three-server reference

v2.7.02026-07-10

-Client attribution: initialize clientInfo is captured and reported in telemetry (see which MCP client makes calls)
-Data fetches send a proper comparedge-mcp User-Agent (were going out as bare node)
-Catalog cache: 6h TTL with serve-stale-on-error for long-lived Claude Desktop processes
-Docs: compare_pricing correctly documented as a prompt; all four prompts now in README
-server.json version unstuck from 2.5.8; version strings unified

v2.6.02026-06-08

-Product count updated to 495
-Fixed serverInfo.version (was incorrectly reporting 2.5.6)
-README: descriptive link anchors, added 556 curated compare pairs note
-glama.json: version bump + updated description

v2.5.82026-05-30

-list_category: added BEHAVIOR, USAGE GUIDELINES, and EXAMPLE QUERIES to tool description - Glama quality score upgraded to A
-GitHub Marketplace listing published: github.com/marketplace/comparedge

v2.5.42026-05-25

-Tool-level telemetry: all 8 tools report call/status/latency to analytics (fire-and-forget, never blocks response)
-Supabase mcp_tool_calls table: per-tool logs with params, response time, ip, user_agent
-track() fn added: POST /api/mcp/track with tool name, params, ok/error status, ms

v2.5.02026-05-22

-4 prompts added: find_best_tool, compare_pricing, evaluate_tool, category_overview
-Fixed stdin close race condition, async tool calls now complete before process exits
-prompts capability now declared in initialize response

v2.4.02026-05-22

-8 tools: added compare_tools, list_category, get_leaderboard, list_categories
-Zero-dependency input validation with self-correction error messages for AI agents
-Published to official modelcontextprotocol/registry (io.github.imkemit-ops/comparedge-mcp)

v2.1.72026-05-17

-Published to official MCP Registry (io.github.imkemit-ops/comparedge-mcp)
-Added GitHub Copilot Chat configuration example
-Added server.json registry metadata

v2.1.52026-05-15

-Migrated data source from Railway API to static JSON endpoints
-Added Glama compatibility: glama.json and Dockerfile
-Added smithery.yaml for Smithery registry indexing

Data by ComparEdge - independent SaaS pricing research

GitHub NPM Package

LLM Cost MCP Server

New@comparedge/llm-cost-mcp · v1.0.0

Estimate and compare the real cost of any LLM workload across 69 models from 17 providers. Live per-token pricing, side-by-side model math, and monthly budget projections without leaving your assistant.

Tools

Prompts

Dependencies

terminal

npx @comparedge/llm-cost-mcp@latest

Publishing to npm shortly. The command resolves the package once it is live; the config below is ready to paste now.

Installation

Runs via npx with no global install. Node.js 18 or later required. Add the block for your client, then restart it.

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "llm-cost": {
      "command": "npx",
      "args": ["-y", "@comparedge/llm-cost-mcp@latest"]
    }
  }
}

Cursor

~/.cursor/mcp.json

{
  "mcpServers": {
    "llm-cost": {
      "command": "npx",
      "args": ["-y", "@comparedge/llm-cost-mcp@latest"]
    }
  }
}

VS Code (GitHub Copilot)

.vscode/mcp.json

{
  "servers": {
    "llm-cost": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@comparedge/llm-cost-mcp@latest"]
    }
  }
}

Tools Reference

The LLM Cost server exposes 6 MCP tools. Each returns structured plain text.

estimate_costtool

Estimate the cost of a workload for a single model from input and output token counts.

Parameters

modelstringrequiredModel id (e.g. claude-opus-4, gpt-4o)

input_tokensnumberrequiredInput (prompt) tokens per call

output_tokensnumberrequiredOutput (completion) tokens per call

callsnumberoptionalNumber of calls to multiply by (default: 1)

Returns

Input cost, output cost, and total for the workload, with the per-million rates used.

compare_models_costtool

Run the same token workload across several models and rank them by cost.

Parameters

modelsstringrequiredComma-separated model ids to compare

input_tokensnumberrequiredInput tokens per call

output_tokensnumberrequiredOutput tokens per call

Returns

Per-model cost table sorted cheapest first, with the delta against the most expensive option.

monthly_budgettool

Project monthly spend for a model from a per-day call volume.

Parameters

modelstringrequiredModel id

input_tokensnumberrequiredInput tokens per call

output_tokensnumberrequiredOutput tokens per call

calls_per_daynumberrequiredExpected calls per day

Returns

Projected daily and monthly cost, plus the token volume the projection is based on.

cheapest_modelstool

Rank the cheapest models for a given token profile.

Parameters

input_tokensnumberrequiredInput tokens per call

output_tokensnumberrequiredOutput tokens per call

limitnumberoptionalNumber of models to return (default: 10)

Returns

Ranked list of the lowest-cost models for that profile with per-call cost and provider.

list_modelstool

List all 69 tracked models with per-token input and output pricing.

Parameters

providerstringoptionalFilter to a single provider slug (e.g. anthropic, openai)

Returns

Model ids with provider, input rate, and output rate per million tokens.

list_providerstool

List all 17 providers with the number of models tracked for each.

Returns

Provider slugs, display names, and model counts.

Prompts Reference

Built-in prompts are guided workflows that chain multiple tools automatically. Select a prompt in your AI client to start a structured session.

estimate_my_workflowprompt

Estimate the LLM cost of a described workflow: token estimates plus per-run and monthly cost.

Arguments

workflowstringrequiredPlain description of the workflow

modelstringoptionalModel name or id

pick_cheapest_modelprompt

Recommend the cheapest model that still meets a stated requirement, with the top candidates priced.

Arguments

taskstringrequiredWhat the model needs to do

forecast_ai_budgetprompt

Project the monthly and yearly bill for a feature from expected volume.

Arguments

modelstringrequiredModel name or id

usagestringrequiredExpected volume

Price Watch MCP Server

New@comparedge/price-watch-mcp · v1.0.0

Track pricing changes across all 495 tools ComparEdge monitors. Surface recent moves, per-product history, category trends, and the biggest increases so your assistant knows when a vendor changed the deal.

Tools

Prompts

Dependencies

terminal

npx @comparedge/price-watch-mcp@latest

Publishing to npm shortly. The command resolves the package once it is live; the config below is ready to paste now.

Installation

Runs via npx with no global install. Node.js 18 or later required. Add the block for your client, then restart it.

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "price-watch": {
      "command": "npx",
      "args": ["-y", "@comparedge/price-watch-mcp@latest"]
    }
  }
}

Cursor

~/.cursor/mcp.json

{
  "mcpServers": {
    "price-watch": {
      "command": "npx",
      "args": ["-y", "@comparedge/price-watch-mcp@latest"]
    }
  }
}

VS Code (GitHub Copilot)

.vscode/mcp.json

{
  "servers": {
    "price-watch": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@comparedge/price-watch-mcp@latest"]
    }
  }
}

Tools Reference

The Price Watch server exposes 5 MCP tools. Each returns structured plain text.

recent_changestool

List the most recent pricing changes across every tracked product.

Parameters

limitnumberoptionalMaximum changes to return (default: 20)

daysnumberoptionalOnly changes within the last N days

Returns

Product, field changed, old and new value, and the date of each change, newest first.

product_changestool

Full price-change history for a single product.

Parameters

slugstringrequiredURL-safe product identifier

Returns

Chronological timeline of every recorded pricing change for that product.

category_changestool

Recent pricing changes within a category.

Parameters

categorystringrequiredCategory slug (e.g. "crm", "llm")

limitnumberoptionalMaximum changes to return (default: 20)

Returns

Recent changes for tools in that category with product, field, and old/new values.

biggest_increasestool

Rank the products with the largest price increases over a window.

Parameters

daysnumberoptionalLook-back window in days (default: 90)

limitnumberoptionalNumber of products to return (default: 10)

Returns

Products ranked by increase, with the amount and percentage change.

check_pricetool

Current verified price and last-changed date for a product.

Parameters

slugstringrequiredURL-safe product identifier

Returns

Current starting price, free tier status, and the date the price last changed.

Prompts Reference

Built-in prompts are guided workflows that chain multiple tools automatically. Select a prompt in your AI client to start a structured session.

recent_price_movesprompt

Summarize the notable SaaS pricing changes in a recent window.

Arguments

daysstringoptionalLook-back window in days

track_productprompt

Current price, free-tier status, and recent direction for one tool.

Arguments

productstringrequiredProduct name or slug

category_price_watchprompt

Recent moves and the sharpest hikes affecting one category.

Arguments

categorystringrequiredCategory slug