WeaverTools is a research platform for the agent architecture developed in the blog series — SPU, Memory System, and Harness as separate cognitive components, integrated through a substrate that operates at hardware speeds rather than network speeds.
This section holds the project’s non-blog material: the hypothesis document that anchors the work, technical reports, design updates, and theoretical pieces that belong to the project rather than to the practitioner-facing series.
What lives here#
- Hypothesis document — the anchoring statement of what WeaverTools is testing and why. (Draft — under active revision.)
- Technical reports — implementation retrospectives, benchmark results, design decisions with their reasoning preserved. (See the current-state snapshots.)
- Updates — periodic notes on where the build sits, what’s working, what’s been deferred.
- Theoretical pieces — material that belongs to the project rather than to the blog series, including drafts that may eventually be promoted to standalone articles.
The blog series presents architectural arguments at a level intended for practitioners. The work here is the substrate of that writing — the engineering and theoretical work that the blog series points at without itself containing.