Durable memory, audit, and continuity for AI coding agents.
An open reference architecture. Free to read, free to use, free to adapt. Six pillars, a small set of conventions, and ~600 lines of glue that turn any AI coding agent into a contractor with a project log instead of a goldfish.
The full one-page treatment. Problem, six pillars, architecture diagrams, hardware reality, comparison to existing tools, cold-start protocol, where this matters.
What Agent OS replaces in the default workflow, and how it sits next to Cursor, Cline, Aider, Continue.dev, LangChain, AutoGen, and the rest.
Memory · Local compute · Vault · Guardrails · Brain backup · Continuity. Each one composable. None of them load-bearing alone.
Hot / warm / cold context tiers, the networked-brain pattern, and how a real session runs from trigger to overnight backup.
A vector database ready to scale to tens of thousands of chunks of your own knowledge. What it is, how to fill it, how to talk to it, and how it's different from /memories/.
What you need at each profile — workstation, mid-range laptop, integrated graphics, phone. Honest about what the GPU buys you.
Three commands sketch the cold-start protocol. Reference implementation links + scheduled-task layout.
Modern AI coding agents are stateless. Every conversation starts cold. Memory is what the host platform decides to keep. Costs are unbounded. Audit is whatever the chat transcript happens to retain. Multi-day campaigns degrade into "we already discussed this — read the scrollback" loops, until eventually the scrollback gets compacted by an opaque summarizer and detail vanishes.
Agent OS treats the agent like a contractor who clocks in every day: they don't remember yesterday's work from their own head, they read the project log. The log lives on disk, in a vector brain, and in immutable cloud storage. The agent's job is to keep that log honest and to read it before acting.