Knowl
- Source
- github.com
- Author
- dat999zx
- Date

Context
Knowl addresses a problem with persistent coding-: an old project fact can remain searchable after a replacement is recorded. It marks replaced facts as superseded, removes them from ordinary and preserves them in history. When replacement is uncertain, both facts remain active until someone resolves the conflict.
The project's own retrieval test isolates that mechanism. Across 100 questions over 455 facts, the current fact appeared first 98% of the time with supersession enabled and 47% with it disabled, using the same ranking method. No generated answers in this test. The result supports retiring obsolete facts as a way to improve retrieval in that setting; it does not measure the correctness of an agent's subsequent work.
- Agent memory — Information an agent can persist and retrieve beyond the immediate context.
- Retrieval — Selecting evidence relevant to the current request.
- Large language model — A model trained on large amounts of data to process and generate language.
- Model Context Protocol (MCP) — A published agreement for an AI application to discover what a separate program offers and ask it to run one of those offerings.
Checking sign-in…
Loading comments…


