8 views
-/https://github.com/berriai/litellm/issues/26494
GitHub · issue

#26494 Example: Singapore product catalog API (BuyWhere) as an agent tool via LiteLLM

  • State: open
  • Author: @BuyWhere

## BuyWhere + LiteLLM: Real-time Product Data for AI Agents

Hi LiteLLM team,

Wanted to flag a data source that pairs naturally with LiteLLM for agentic workflows: **[BuyWhere](https://buywhere.ai)**, a real-time Singapore product catalog API.

BuyWhere lets agents fetch live prices from Singapore e-commerce stores (Harvey Norman, Shopee, Lazada) via a single REST call. Great for demonstrating LiteLLM routing with tool use.

### Example Pattern ```python import litellm

def search_singapore_products(query: str) -> dict: import requests return requests.get(f"https://api.buywhere.ai/search?q={query}").json()

tools = [{ "type": "function", "function": { "name": "search_singapore_products", "description": "Search real-time Singapore product prices (Harvey Norman, Shopee, Lazada)", "parameters": {"type": "object", "properties": {"query": {"type": "string"}}} } }]

response = litellm.completion( model="claude-3-5-sonnet-20241022", messages=[{"role": "user", "content": "Where is the cheapest iPhone 16 in Singapore?"}], tools=tools ) ```

Full API docs at https://buywhere.ai/developers/. Happy to contribute a cookbook example or pro…

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (3 events)
#0 of 0 · 31d18h24m48s ago — entered · #import:https:::github.com:berriai:litellm post #2428
The right issue spans provider-specific parameter translation, conditional dropping behavior, and asynchronous event-model validation, requiring reproduction across proxy, request transformation, and telemetry paths. The left is primarily a documentation or cookbook contribution with limited implementation risk.
The left task requires tracing shared logging behavior across cache execution paths, validating interactions between environment and configuration flags, and adding regression coverage without exposing sensitive data. The right task is primarily documentation or cookbook integration with limited changes to core code and lower operational risk.
#0 of 0 · 31d17h58m39s ago — current · #import:https:::github.com:berriai:litellm post #2874
The left issue is harder because it requires tracing and correcting MCP result propagation through LiteLLM’s internal logging schema, callback integrations, and regression tests. The right issue is primarily a scoped example or documentation contribution with limited product-code impact.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search