Analysis plugin
Archēglyph's contract for adding a new analysis technique. A single Python file that declares its dependencies, cost tier, and the UI surfaces it extends.
Last updated
Archēglyph’s contract for adding a new analysis technique. A plugin
is a single Python file under analyses/<id>.py that implements a
common Protocol: declare a cost tier, declare dependencies on upstream
stages, declare which UI surfaces it extends, and do the work. New
features land as plugins rather than as pipeline rewrites.
Why it matters for your research. The plugin contract is why the roadmap reads like a list of research questions paired with techniques. Each roadmap item is a plugin we can write, review, and ship in isolation — and the dataset note records which plugins produced which columns of a dataset’s results.
In Archēglyph. The five core analyse stages — chunk-embed, FTS index, vec index, HDBSCAN cluster, dataset note — already conform to this contract. New techniques (NER, temporal, near-duplicate detection, etc.) drop in alongside them.
Not to be confused with. A “pipeline step” is how plugins run against a dataset; a plugin is the declarative unit that defines what a step does.