Every agent tutorial ends with "wire up your tools." Nobody tells you that means managing auth for 10 different APIs, writing retry logic, handling rate limits, and keeping tool schemas in sync with your LLM's function calling format. Here's the shortcut.
If you've tried building an agent with real capabilities, you've hit this list:
That's the infrastructure tax before you've written any actual agent logic. Slopshop is the tool layer that handles all of it.
Post a task in plain English. The agent picks the right tools, chains them, and returns a result. You don't define any tool schemas.
curl -X POST https://slopshop.gg/v1/agent/run \ -H "Authorization: Bearer demo_key_slopshop" \ -H "Content-Type: application/json" \ -d '{ "task": "Check the SSL certificate for stripe.com and tell me when it expires", "max_steps": 3 }'
{
"ok": true,
"result": "stripe.com SSL cert expires 2026-08-14. Issuer: DigiCert. 141 days remaining.",
"steps": [
{"tool": "ssl_check", "input": "stripe.com", "credits": 3}
],
"credits_used": 3,
"_engine": "real"
}
Not sure which tools exist? The quickstart endpoint returns everything organized by category with credit costs.
curl https://slopshop.gg/v1/quickstart \
-H "Authorization: Bearer demo_key_slopshop"
{
"tools_available": 422,
"categories": {
"network": {
"count": 28,
"tools": ["dns_lookup", "ssl_check", "http_headers", "whois", ...]
},
"code_execution": {
"count": 4,
"tools": ["exec_python", "exec_js", "exec_bash", "exec_sql"]
},
"memory": {
"count": 6,
"tools": ["memory_set", "memory_get", "memory_search", ...],
"note": "free tier — 0 credits"
},
"text": {
"count": 61,
"tools": ["summarize", "sentiment", "extract_entities", "translate", ...]
}
},
"free_credits": 2000
}
If you're building with Claude (Code, Desktop, or API), the Slopshop MCP server wires the full catalog of tools directly into Claude's tool-use protocol. Add this to your Claude config:
{
"mcpServers": {
"slopshop": {
"command": "npx",
"args": ["slopshop-mcp"],
"env": {
"SLOPSHOP_API_KEY": "your_key_here"
}
}
}
}
Claude now has access to the full catalog of tools as native MCP tools. It can call ssl_check, exec_python, memory_set, or any other tool by name — with full observability on your end.
Use Slopshop as a backend for OpenAI function calling. Define the tool once, route the call to Slopshop, get back a validated result.
# Define the Slopshop tool for GPT tools = [{ "type": "function", "function": { "name": "slopshop_run", "description": "Run any of the real tools — network, code, data, memory", "parameters": { "type": "object", "properties": { "tool": {"type": "string", "description": "Tool slug, e.g. ssl_check"}, "input": {"type": "object", "description": "Tool-specific parameters"} }, "required": ["tool", "input"] } } }] # When GPT calls the function, route to Slopshop def handle_tool_call(tool_name, tool_input): res = requests.post( "https://slopshop.gg/v1/tools/run", headers={"Authorization": f"Bearer {SLOPSHOP_KEY}"}, json={"tool": tool_name, "input": tool_input} ) return res.json()["output"]
The docs include complete schemas for every tool in both OpenAI and Anthropic formats, plus a schema-generation endpoint that builds the function definition automatically.
One API key. 500 free credits. No setup beyond npm install.