Skip to content

SearchForge v0.1.0 — open web search for every RAG stack

Choose a tag to compare

View all tags
@divyanshu-iitian divyanshu-iitian released this 23 Jul 13:11
· 12 commits to main since this release
v0.1.0
fd0b20d

SearchForge gives LLMs, agents, and RAG pipelines a provider-neutral web search layer without requiring another LLM.

Highlights

  • SearXNG-first, self-hosted full-web metasearch
  • Optional Brave Search adapter and no-key Wikipedia fallback
  • REST API, MCP stdio tool, CLI, and TypeScript SDK
  • Concurrent provider calls with independent timeouts and partial-failure recovery
  • Canonical URL deduplication and Reciprocal Rank Fusion
  • Citation-ready URLs, snippets, provenance, and provider health
  • Bounded TTL cache, optional API-key auth, rate limiting, and request IDs
  • Docker Compose stack with JSON-enabled SearXNG
  • OpenAPI 3.1 contract
  • Zero known dependency vulnerabilities
  • Tested on Node.js 20, 22, and 24

Start

git clone https://github.com/divyanshu-iitian/SearchForge.git
cd SearchForge
docker compose up --build
curl -s http://localhost:3000/v1/search \
  -H "content-type: application/json" \
  -d '{"query":"open source vector databases","limit":5}'

Search results are untrusted external evidence. Preserve citations and apply prompt-injection defenses in downstream RAG applications.

Assets 2
Loading