Built a repo-backed system that helps agents use Brandon's writing, decisions, and source material instead of guessing from a prompt.
- role
- system designer / builder
- context
- Writing, source material, and agent memory
evidence
- Thought catalog reframed as a source of working context
- Website copy generated from explicit source material
- CLI and ontology direction for agents that need to inspect the source map
The thought catalog is becoming more than a place to store documents. It is meant to help agents find the writing, decisions, and source material they need before they write or design on Brandon's behalf.
The practical loop is simple: keep source material in the catalog, generate website-ready copy from it, and let agents inspect the source map through tools instead of guessing from a single prompt.
The product bet is that agents get better when they can use durable source material and explicit decisions, not just a style prompt. The deeper implementation can use ontology and NLP tools, but the public value is better context for better output.