tech-search

Automate deep tech research by routing queries through a structured pipeline of decomposition, parallel search, evaluation, and synthesis to produce documented findings. It handles software engineering topics like architectures, libraries, and benchmarks.

Query decomposition in a tech research pipeline breaks down a complex user query into smaller parts to enable parallel Haiku-based web searches. This structured phase ensures comprehensive sourcing of authoritative data before final synthesis and documentation.

No external dependencies are required to run this structured tech research workflow. The pipeline is self-contained and performs parallel search, evaluation, and synthesis without external setup, saving reproducible results directly under the docs/research directory.

Yes, you can preserve exact code examples when generating technical documentation. The research pipeline applies synthesis rules that maintain precise data and code snippets from authoritative sources directly into the final documented findings.

The best way to source authoritative data for library benchmarks is using a self-contained research pipeline that executes parallel web searches and structured evaluation. It synthesizes precise benchmark data into reproducible documentation files.

Parallel search improves software engineering knowledge discovery by simultaneously querying decomposed topics to gather diverse authoritative sources. This phase-based workflow accelerates data collection before synthesizing the evaluated findings into structured documentation.