AgentShare Price API
The AgentShare Price API lets AI agents search and compare structured marketplace prices and offers via MCP (stdio or Streamable HTTP) or direct REST calls.
price_search: Search across connected marketplaces using free-text queries (product name, model, brand, or category) to get multiple structured offers with prices, sources, and freshness metadata. Supports optional result limits (1–50).best_offer: Get the single cheapest in-stock offer for a product query.best_offer_under_budget: Find the best offer constrained by a specified maximum price (e.g. "under $X").service_meta: Retrieve API capabilities, rate limits, and integration hints. Safe to call without an API key — useful for onboarding or pre-flight checks.
Additional details:
Results include freshness metadata (
crawled_at,data_age_seconds,freshness_status) for reliability assessment.Current marketplace coverage available via
GET /coverage.Requires an API key (free tier: 100 requests/month).
Provides access to structured marketplace prices and offers from AliExpress, enabling AI agents to retrieve product data, pricing information, and availability with freshness metadata for reliability assessment.
Provides access to structured marketplace prices and offers from Amazon, enabling AI agents to retrieve product data, pricing information, and availability with freshness metadata for reliability assessment.
Provides access to structured marketplace prices and offers from eBay, enabling AI agents to retrieve product data, pricing information, and availability with freshness metadata for reliability assessment.
AgentShare
Structured marketplace prices and offers for AI agents — REST API and MCP (Streamable HTTP).
Site & docs | |
MCP endpoint |
|
Discovery |
|
This repository holds the stdio MCP server used with Claude Desktop, Cursor, and other MCP clients, plus minimal REST examples. The live API and full integration guides are on the site above.
For AI agents (machine readable)
AGENTS.md— mission, tool names, auth, response envelope, copy-paste MCP JSON, and links to OpenAPI.llms.txt— same essentials in a short, crawl-friendly single file (for LLM / GEO workflows).openapi.json— OpenAPI 3.0 spec for the REST surface used by this MCP (/api/v1/search,/offers/*,/meta). The full production spec (all routes) is always at https://agentshare.dev/openapi.json (single source of truth; no duplicate docs repo).mcp-config.json— ready-to-pastemcpServersblock fornpx mcp-remote(Claude Desktop / Cursor) pointing at the remote Streamable HTTP endpoint.
Getting an API key
To use this MCP server or the REST API, you need an API key. Visit https://agentshare.dev/pricing to get your free tier key (100 requests / month on the public free plan at time of writing — always confirm on the site).
For humans: quick install (pip)
git clone https://github.com/anhmtk/agentshare-mcp.git
cd agentshare-mcp
pip install -r integrations/mcp_server/requirements.txt
export API_KEY=your_api_key # Windows: $env:API_KEY="..."
# optional: export BASE_URL=https://agentshare.dev
python integrations/mcp_server/server.pyGet a key: https://agentshare.dev/pricing
Quick install — Claude Desktop (claude_desktop_config.json)
Use a local stdio server (Python). Replace the path with the absolute path to server.py in your clone of this repo.
{
"mcpServers": {
"agentshare": {
"command": "python",
"args": ["/ABSOLUTE/PATH/TO/agentshare-mcp/integrations/mcp_server/server.py"],
"env": {
"API_KEY": "your-api-key-here",
"BASE_URL": "https://agentshare.dev"
}
}
}
}Get a key: https://agentshare.dev/pricing
On Windows, prefer forward slashes in
args, e.g.D:/code/agentshare-mcp/integrations/mcp_server/server.py.
Remote MCP (Streamable HTTP): clients that support URL + API key headers can use https://agentshare.dev/mcp/ with X-API-Key or Authorization: Bearer. Use mcp-config.json for npx mcp-remote. Details: MCP Quickstart (section MCP).
Advanced — HTTP via mcp-remote (Node / npx): if you use mcp-remote to bridge HTTPS → stdio, pass your key with --header (see troubleshooting in Cursor MCP setup); this repo does not publish an npx agentshare-mcp package.
Clone & run the stdio MCP locally
From the repo root:
pip install -r integrations/mcp_server/requirements.txt
export API_KEY=your_api_key
# optional: export BASE_URL=http://localhost:8000
python integrations/mcp_server/server.pySame as python integrations/mcp_server/run.py. See integrations/mcp_server/README.md for tools and env.
Code examples (REST)
The examples/ folder has small scripts that call the public JSON API (GET /api/v1/search, …). Useful for testing a key before wiring MCP.
MCP directories & listings
Production discovery (no auth) — use these URLs when submitting to catalogs:
Resource | URL |
| |
| |
Server card (Smithery-style) |
Third-party indexes (search for AgentShare or your listing URL):
License
MIT — see LICENSE.
GitHub — CI & repo traffic
CI: validates MCP package imports on push/PR — Actions → CI.
Traffic: GitHub shows aggregate views/clones only (Insights → Traffic for maintainers). Weekly GitHub traffic snapshot writes API results to the run Summary (same idea as Insights).
Scope (honest & forward-looking)
AgentShare aggregates product and offer data from connected marketplaces and affiliate sources. Coverage and freshness vary by source; the API returns freshness metadata (e.g. crawled_at, data_age_seconds, freshness_status) so agents can judge reliability.
Direction: expand toward global e‑commerce and major international marketplaces as integrations mature. For the current focus, see GET /coverage: https://agentshare.dev/coverage
Legal
Terms and privacy: https://agentshare.dev/terms · https://agentshare.dev/privacy
Latest Blog Posts
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/anhmtk/agentshare-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server