A Stunspot Augment

Yammerknit

Build thoughtbases from transcripts and documents: reusable cognitive architecture for new comparisons, reframes and reasoning.

Codex and Claude packages Source-preserving workspaces Optional YouTube subtitles

What it does

Meaning gets structure. Sources keep custody.

Yammerknit follows semantic gravity instead of transcript order. It builds meaning-nodes, maps their latent functions and connects them through bridges that let a later model think with the source in a new situation.

Shape the source

Create a canonical Yammerknit from supplied transcripts or documents, with the complete cluster mesh, bridges, prompt seeds, and sigil lexicon.

Keep the evidence

Separate original bytes, normalized text, artifacts, revisions, hashes, and manifests when you choose a workspace.

Prepare for vector RAG

Produce a semantic artifact for a later embedding and vector-storage pipeline—without pretending the storage layer is already there.

A governed method

Not summary sludge in a clever hat.

Sam Walker’s canonical SOP governs thoughtbase construction, with Rupert Giles as document guardian and information architect. Sigil stacks, echo shards and the keystone relational mesh carry the method; source custody keeps factual claims and proposed transfers distinguishable.

Clear boundary: Yammerknit creates retrieval-ready semantic meshes. It does not include an embedding model, vector database, ingestion service, or retrieval runtime.
An archival document unfolding into connected idea clusters, cross-links, and retrieval nodes.

Shortest path to value

One kit. Your host. Then talk normally.

Download once

The customer ZIP contains the Codex plugin, Claude skill, documentation, branding, and package custody records.

Install for your agent

Use the included host package. Other capable agents can use the complete self-contained skill folder.

Supply a source

Ask: “Build a Yammerknit thoughtbase from this transcript. Construct the keystone bridges and seed reasoning I can reuse.”