Summary of an essay on muthub.org
Read the full essay on muthub.org →
Reviewed by Muthu · Essay · Last verified Oct 18, 2026 · Drafted Oct 17, 2026 by seed-import/1.0 · Re-check by Oct 17, 2027 · 1 source
Overview
The tempting answer to fragmented context is to pour everything into one place. We tried that with data, and many lakes became swamps. A context layer has to be a curated overlay: retrieval-first, owned, and pointing back to the systems that hold the truth.1
What it covers
- Why bulk ingestion fails for context
- A curated overlay instead
- Collection versus curation
- Signs you are building a context swamp
- Retrieval that respects policy
- Governance debt, applied to context
- Isn’t this just a knowledge graph?
Bottom line
We already know what happens when collection is mistaken for knowledge. Build the context layer as a curated overlay: small, owned, retrieval-first and pointing back to the source of truth. Otherwise you are building the next swamp, with agents swimming in it.
Sources
- Not another data lake: context needs curation, not collection · published Jul 21, 2026
See also
- Context fragmentation: why your agents keep starting from zero - The knowledge an agent needs is scattered across code, tickets, chat, wikis, incident tools and people’s heads. Each fragment is accurate and useless alone. Fragmentation, more than model quality, is what makes enterprise agents brittle.
- Passive capture: knowledge that writes itself, with humans who approve it - Documentation fails because it is a second job. The knowledge we need is already produced as a by-product of work: commits, tickets, incidents, reviews and decisions. Capture it passively, distill it, and keep a human approval gate before anything becomes trusted.
Footnotes
-
Not another data lake: context needs curation, not collection, muthub.org, 2026-07-21. ↩