Engram
My work on giving agents a durable memory: what to keep, what to let go, and how to recall it at the right moment.
My research
Named research I keep coming back to, aimed at real production problems in autonomous systems, not theory. Agent memory, intent graphs, goals and contracts. One line on what each is for.
§ 04.1REFERENCES
iCPG
An intent-augmented code property graph. It tracks not just what the code does but why it exists, so agents can reason about change safely.
Mnemos
Task-scoped memory that survives context compaction by treating facts, decisions and code references as distinct, separately governed things.
Telos
Formal goals and contracts for autonomous work, so an agent can check its output against the intent it was given.
DAE
Part of the same body of named research, aimed at a real production problem in autonomous systems rather than theory. I am still writing this one up.
Lexon
Another strand in the same line of work on autonomous systems, grounded in a production problem, not a paper. Notes are in progress.