web-search-plus-mcp 1.2 release notes

Version 1.2 packages the portable Web Search Plus 3.2 provider contract for MCP
clients and adds Hound as an optional local Search and Extract provider.

Highlights

• 13 source-result Search providers and 9 Extract providers
• optional local Hound sidecar through MCP Streamable HTTP
• Hound remains explicit-only until auto_allow is enabled
• loopback-only endpoint validation with no redirects or proxy inheritance
• Hound response caching disabled so Web Search Plus retains cache ownership
• full Hound Search and Extract projection through the existing two MCP tools
• mcp>=1.26,<2 and explicit httpx>=0.27,<1 runtime dependencies

Stable public surface

The server still exposes exactly two tools:

• web_search
• web_extract

Hound is additive in both provider enums. Existing tool names, source-only
contracts, typed errors, evidence receipts, bounded extraction, and legacy
result fields remain stable.

Hound operation

Hound is installed and run separately. It is not embedded in the wheel. The
adapter accepts only literal loopback endpoints such as:

http://127.0.0.1:8765/mcp
http://[::1]:8765/mcp

See HOUND.md for installation, configuration, keyless trade-offs,
security boundaries, and opt-in automatic routing.

Attribution

Hound / Master Fetch is an
independent MIT-licensed project created and maintained by
Bishesh Bhandari (dondai1234). Web Search Plus
connects to it as a separately installed sidecar; it does not vendor or fork the
project.

The Hound integration was validated against hound-mcp 11.1.6.

Upgrade

pip install --upgrade "web-search-plus-mcp==1.2.0"

Or pin the version in an MCP host configuration:

{
  "command": "uvx",
  "args": ["web-search-plus-mcp==1.2.0"]
}

No migration is required for users who do not configure Hound.