feat(search): add TinyFish as search provider

[simantak-dabhade]

Summary

Adds TinyFish as the 14th search provider in LiteLLM, following the same GET-based pattern as Brave/Serper.

• Implements TinyfishSearchConfig(BaseSearchConfig) with full request/response transformation
• Maps unified LiteLLM params (country → location, search_domain_filter → site: operators)
• Auth via X-API-Key header (from api_key param or TINYFISH_API_KEY env var)
• 6 unit tests covering: basic search, country mapping, domain filter injection, language passthrough, empty results, missing API key

Files changed (8)

File	Change
litellm/llms/tinyfish/search/__init__.py	New module export
litellm/llms/tinyfish/search/transformation.py	Core TinyfishSearchConfig implementation
litellm/types/utils.py	Add TINYFISH to SearchProviders enum
litellm/utils.py	Register config in get_provider_search_config()
model_prices_and_context_window.json	Add tinyfish/search pricing entry
provider_endpoints_support.json	Add tinyfish endpoint metadata
tests/code_coverage_tests/enforce_llms_folder_style.py	Add tinyfish to SEARCH_PROVIDERS
tests/search_tests/test_tinyfish_search.py	6 unit tests

Test plan

• uv run pytest tests/search_tests/test_tinyfish_search.py -v — all 6 tests pass
• Lint passes (make lint-ruff)
• Black formatting applied

🤖 Generated with Claude Code

[greptile-apps]

Greptile Summary

This PR adds TinyFish as a 14th search provider in LiteLLM, following the established GET-based pattern of Brave and Serper. The implementation is clean and well-structured, with correct parameter mappings (country → location, max_results capped and applied client-side) and proper auth via X-API-Key.

• TinyfishSearchConfig correctly extends BaseSearchConfig, encodes query params via urlencode in get_complete_url, and uses the same _tinyfish_params wrapper dict pattern as Brave uses for _brave_params.
• Six unit tests cover all major code paths using properly mocked HTTP calls — no real network requests.
• Plumbing changes (enum, provider map, pricing JSON, folder style test) are minimal and consistent with how other search-only providers are registered.

Confidence Score: 5/5

This is a self-contained new provider addition with no changes to existing code paths; all required plumbing is in place and follows established patterns.

The implementation is additive-only, uses LiteLLM's existing HTTP handler infrastructure, and mirrors the Brave/Serper GET-provider pattern faithfully. The two observations are cosmetic style differences, not functional defects.

No files require special attention.

Important Files Changed

Filename	Overview
litellm/llms/tinyfish/search/transformation.py	Core TinyFish search config — follows the GET-based Brave/Serper pattern; explicit mappings for country→location and max_results; passthrough loop for provider-specific extras
litellm/llms/tinyfish/search/init.py	Minimal module init exporting TinyfishSearchConfig — no issues
litellm/types/utils.py	Adds TINYFISH to SearchProviders enum following existing pattern
litellm/utils.py	Registers TinyfishSearchConfig in get_provider_search_config() alongside other search providers
model_prices_and_context_window.json	Adds tinyfish/search pricing entry with input_cost_per_query=0.0 and subscription-based pricing note
provider_endpoints_support.json	Adds tinyfish provider metadata; the docs URL points to a page that doesn't exist yet
tests/code_coverage_tests/enforce_llms_folder_style.py	Adds tinyfish to SEARCH_PROVIDERS allowlist — correct placement to pass the llms folder style check
tests/search_tests/test_tinyfish_search.py	6 unit tests with properly mocked HTTP calls; covers basic search, country mapping, domain filter injection, language passthrough, max_results truncation, empty results, and missing API key

_{Reviews (2): Last reviewed commit: "fix(search/tinyfish): handle max_results..." | Re-trigger Greptile}

[greptile-apps]

The max_results unified parameter is a member of get_supported_perplexity_optional_params(), which means it is intentionally excluded from the passthrough loop. But there is no explicit mapping of max_results to a TinyFish-specific parameter, so passing max_results=5 to litellm.asearch(…, search_provider="tinyfish") is silently swallowed — the API receives no limit hint. Every other provider that supports result limiting (Brave → count, Serper → num) has an explicit mapping; TinyFish needs the same treatment once the correct API parameter name is confirmed from the TinyFish docs.

Suggested change

	request_data: TinyfishSearchRequest = {"query": query}
	if "country" in optional_params:
	request_data["location"] = optional_params["country"]
	request_data: TinyfishSearchRequest = {"query": query}
	if "max_results" in optional_params:
	# TODO: replace "num_results" with the actual TinyFish API parameter name
	request_data["num_results"] = optional_params["max_results"]
	if "country" in optional_params:
	request_data["location"] = optional_params["country"]

[simantak-dabhade]

Fixed in b251f35 — max_results is now mapped in transform_search_request and used for client-side truncation in transform_search_response (reading it back from the URL query params, same pattern as Brave). Added a dedicated test as well.

[greptile-apps]

include_thumbnail is an include/exclude flag and should be typed as bool. Keeping it as str is a type-checking gap — static analysis tools will not catch callers passing a non-string value, and urlencode will serialize a Python bool as "True"/"False" (capital T/F) rather than the lowercased "true"/"false" most REST APIs expect.

Suggested change

	class TinyfishSearchRequest(_TinyfishSearchRequestRequired, total=False):
	location: str
	language: str
	page: int
	include_thumbnail: str
	class TinyfishSearchRequest(_TinyfishSearchRequestRequired, total=False):
	location: str
	language: str
	page: int
	include_thumbnail: bool

[simantak-dabhade]

Fixed in b251f35 — changed to bool.

[simantak-dabhade]

Hey team — looking for a review on this PR when you get a chance!

This adds TinyFish as a new search provider (the 14th), following the same GET-based pattern as Brave/Serper. Clean 8-file diff, 341 additions.

Update: Pushed a second commit addressing Greptile's feedback:

• Added client-side max_results truncation (TinyFish API doesn't have a count param, so we handle it like Brave does — pass it as a query param and truncate in transform_search_response)
• Fixed include_thumbnail type annotation (bool not str)
• Added a 7th test for max_results truncation

cc @ishaan-berri @yuneng-berri

[simantak-dabhade]

@greptileai review

[oss-pr-review-agent-shin]

🤖 litellm-agent: This PR is currently BLOCKED from merge.

Score: 3/5 ❌

Why blocked:

• 1 PR-related CI failure (This PR will be auto-closed as it lacks a screenshot for proof of fix. Please include one in the PR description. Add the screenshot-exempt label if this PR has no visible output (e.g. pure docs, CI config).) (pr_related_failures, -2 pts)

Details: Score docked for: 1 PR-related CI failure (This PR will be auto-closed as it lacks a screenshot for proof of fix. Please include one in the PR description. Add the screenshot-exempt label if this PR has no visible output (e.g. pure docs, CI config).).

Fix the issues above and push an update — the bot will re-review automatically.

Note: This bot is still in beta and might not always work as expected. Please share any feedback via Slack.