The changelog tells you what shipped. It does not tell you why a
ranking signal was added, which three hypotheses died before the fourth one
held, or what the numbers looked like before and after. That is what this blog
is for.
What goes here
TeaRAGs is an MCP server for code search that enriches every retrieved chunk
with git history — authorship, churn, bug-fix rate, ownership — so an agent
ranks stable, owned code above whatever merely looks similar. Most of the
interesting work behind that sentence is invisible in a release note:
- Retrieval — why a rerank preset weights the signals it does, and what
happens to result quality when the weights are wrong.
- Codegraph — symbol resolution against real corpora, where the resolver
chain gives up, and what the residual gap actually consists of.
- Performance — indexing throughput, profiling runs, measured wins and the
optimizations that turned out to be noise.
- Agents — how a coding agent consumes this data, and where the tooling
still hands it the wrong thing.
Posts here are measurement-first. A claim about ranking or throughput arrives
with the corpus it was measured on and the number it moved.
Where the rest lives
- Documentation — installation, tools, configuration, architecture.
- Changelog — every released version, generated from commits.
- GitHub — source, issues, releases.
The blog has an RSS feed
and an Atom feed if you
would rather not check back.
]]>