Wasm Functions
🤖 What Are Tool Calls in LLMs?
Tool calls in Large Language Models (LLMs) allow the model to interact with external tools, APIs, or functions to fetch real-world data or perform tasks beyond its native capabilities.
Instead of relying solely on pre-trained knowledge, LLMs can dynamically query functions like get_flight_times to provide accurate, real-time information.
🔹 How Do Tool Calls Work?
When an LLM identifies a request that matches a tool’s function, it formats the request into a structured call (like an API request) and processes the response before delivering it to the user.
✈️ What is get_flight_times?
The get_flight_times function is a tool call that helps you retrieve flight durations between two locations. By providing a departure and arrival destination, you can get accurate estimates of how long your flight will take.
🛠️ How Does it Work?
The function operates by taking in two key parameters:
- departure 🛫 – The airport or city where your flight begins.
- arrival 🛬 – The airport or city where your flight ends.
Once you provide these details, get_flight_times fetches the estimated duration of direct and indirect flights between the specified locations. The function considers factors like average flight speeds, distances between airports, and typical air traffic conditions.
📝 Example Usage
Imagine you want to check the flight duration from New York (JFK) to Los Angeles (LAX):
get_flight_times(departure="JFK", arrival="LAX")
The function will return the estimated flight duration, such as 5h 30m for a direct flight or longer if layovers are involved.
Next flight function example
In this example, we will create a tool call function named next_flight that will take two IATA codes (airport codes like CDG for Paris Charles-De-Gaulle) and will return the flight number and when the flight will take off from now.
Function configuration
'inline module';
exports.tool_call = function(args) {
const formattedArguments = JSON.parse(args);
return JSON.stringify({
nextflight: `flight ${parseInt(Math.random() * 10000)} from ${formattedArguments["departure"]} to ${formattedArguments["arrival"]} is taking off in ${parseInt(Math.random() * 10)} hours`
})
}
Function parameters
{
"departure": {
"type": "string",
"description": "The departure airport (IATA airport Code)"
},
"arrival": {
"type": "string",
"description": "The arrival airport (IATA airport Code)"
}
}
Request to our endpoint
curl --request POST \
--url http://wasm-tool-function.oto.tools:8080/ \
--header 'content-type: application/json' \
--data '{
"messages": [
{
"role": "user",
"content": "Could you give me the next flight from CDG to DXB please ?"
}
]
}'
LLM Response with our tool function next_flight
{
"id": "chatcmpl-FzOsLwlxF7cgTkpBfSPfhYEEvw53X274",
"object": "chat.completion",
"created": 1743519776,
"model": "gpt-4o-mini",
"system_fingerprint": "fp-Jz72gmHTIAQw2xYamDRy3Q8kMSPUsMb0",
"choices": [
{
"index": 0,
"message": {
"role": "assistant",
"content": "The next flight from CDG (Charles de Gaulle Airport) to DXB (Dubai International Airport) is flight 3755, which is taking off in 9 hours."
},
"logprobs": null,
"finish_reason": "stop"
}
],
"usage": {
"prompt_tokens": 209,
"completion_tokens": 37,
"total_tokens": 246,
"completion_tokens_details": {
"reasoning_tokens": 0
}
}
}
Full entity configuration
{
"_loc": {
"tenant": "default",
"teams": [
"default"
]
},
"id": "tool-function_f6562221-1c0f-4f8e-aae6-279beb0144dd",
"name": "next_flight",
"description": "This function returns the next flight.",
"metadata": {},
"tags": [],
"strict": true,
"parameters": {
"departure": {
"type": "string",
"description": "The departure airport (IATA airport Code)"
},
"arrival": {
"type": "string",
"description": "The arrival airport (IATA airport Code)"
}
},
"required": null,
"backend": {
"kind": "QuickJs",
"options": {
"jsPath": "'inline module';\n\nexports.tool_call = function(args) {\n const formattedArguments = JSON.parse(args);\n return JSON.stringify({\n nextflight: `flight ${parseInt(Math.random() * 10000)} from ${formattedArguments[\"departure\"]} to ${formattedArguments[\"arrival\"]} is taking off in ${parseInt(Math.random() * 10)} hours`\n })\n}"
}
},
"kind": "ai-gateway.extensions.cloud-apim.com/ToolFunction"
}
Provider configuration
{
"_loc": {
"tenant": "default",
"teams": [
"default"
]
},
"id": "provider_6699f127-34d1-4fef-b3ad-6bcf592ed9c6",
"name": "OpenAI provider + tool wasm function",
"description": "An OpenAI LLM api provider",
"metadata": {},
"tags": [],
"provider": "openai",
"connection": {
"base_url": "https://api.openai.com/v1",
"token": "${vault://local/openai-token}",
"timeout": 30000
},
"options": {
"model": "gpt-4o-mini",
"frequency_penalty": null,
"logit_bias": null,
"logprobs": null,
"top_logprobs": null,
"max_tokens": null,
"n": 1,
"presence_penalty": null,
"response_format": null,
"seed": null,
"stop": null,
"stream": false,
"temperature": 1,
"top_p": 1,
"tools": null,
"tool_choice": null,
"user": null,
"wasm_tools": [
"tool-function_f6562221-1c0f-4f8e-aae6-279beb0144dd"
],
"mcp_connectors": [],
"allow_config_override": true
},
"provider_fallback": null,
"context": {
"default": null,
"contexts": []
},
"models": {
"include": [],
"exclude": []
},
"guardrails": [],
"guardrails_fail_on_deny": false,
"cache": {
"strategy": "none",
"ttl": 300000,
"score": 0.8
},
"kind": "ai-gateway.extensions.cloud-apim.com/Provider"
}
Endpoint configuration
{
"_loc": {
"tenant": "default",
"teams": [
"default"
]
},
"id": "route_52b9e4b69-d161-4e26-9a32-4bcc4c25e2db",
"name": "wasm tool function example",
"description": "wasm tool function example",
"tags": [],
"metadata": {},
"enabled": true,
"debug_flow": false,
"export_reporting": false,
"capture": false,
"groups": [
"default"
],
"bound_listeners": [],
"frontend": {
"domains": [
"wasm-tool-function.oto.tools"
],
"strip_path": true,
"exact": false,
"headers": {},
"query": {},
"methods": []
},
"backend": {
"targets": [
{
"id": "target_1",
"hostname": "request.otoroshi.io",
"port": 443,
"tls": true,
"weight": 1,
"predicate": {
"type": "AlwaysMatch"
},
"protocol": "HTTP/1.1",
"ip_address": null,
"tls_config": {
"certs": [],
"trusted_certs": [],
"enabled": false,
"loose": false,
"trust_all": false
}
}
],
"root": "/",
"rewrite": false,
"load_balancing": {
"type": "RoundRobin"
},
"client": {
"retries": 1,
"max_errors": 20,
"retry_initial_delay": 50,
"backoff_factor": 2,
"call_timeout": 30000,
"call_and_stream_timeout": 120000,
"connection_timeout": 10000,
"idle_timeout": 60000,
"global_timeout": 30000,
"sample_interval": 2000,
"proxy": {},
"custom_timeouts": [],
"cache_connection_settings": {
"enabled": false,
"queue_size": 2048
}
},
"health_check": {
"enabled": false,
"url": "",
"timeout": 5000,
"healthyStatuses": [],
"unhealthyStatuses": []
}
},
"backend_ref": null,
"plugins": [
{
"enabled": true,
"debug": false,
"plugin": "cp:otoroshi.next.plugins.OverrideHost",
"include": [],
"exclude": [],
"config": {},
"bound_listeners": [],
"plugin_index": {
"transform_request": 0
}
},
{
"enabled": true,
"debug": false,
"plugin": "cp:otoroshi_plugins.com.cloud.apim.otoroshi.extensions.aigateway.plugins.OpenAiCompatProxy",
"include": [],
"exclude": [],
"config": {
"refs": [
"provider_6699f127-34d1-4fef-b3ad-6bcf592ed9c6"
]
},
"bound_listeners": [],
"plugin_index": {}
}
],
"kind": "proxy.otoroshi.io/Route"
}