ICM workspaces
Agents do their best work when they are handed the right context for the step they are on — and people can see exactly what that context was. MemoryRelay is building a hosted home for that context, on the Interpretable Context Methodology.
The method
ICM keeps what an agent needs to know in a small workspace of plain files. Instead of searching everything and hoping the right pieces come back, the agent follows the workspace's structure to the one stage it is working on.
Folders carry sequencing
Work is split into ordered stages. Each stage is a folder with a contract that says what it takes in, what it produces, and who checks it.
Hierarchy scopes context
An agent walks from one short entry file to the stage it needs, and loads only what that stage lists — not everything you have ever written down.
Plain files carry state
Context, decisions and outputs are ordinary files. A person can read them, review them and keep them in git; an agent can walk them with no service running.
release-notes/├── CLAUDE.md # entry: where to go├── CONTEXT.md # always-true context└── stages/ ├── 01-collect/ │ ├── CONTEXT.md # stage contract │ └── output/ ├── 02-draft/ │ ├── CONTEXT.md │ └── output/ └── 03-review/ ├── CONTEXT.md └── output/Illustrative only. Workspaces take several shapes — a pipeline of stages is one of them.
What MemoryRelay is building
The method works with nothing but a folder. MemoryRelay will host, version and deliver those workspaces for teams and their agents.
Versioned releases
Publish a workspace as an immutable release, so every agent run knows exactly which context it was given.
Stage-scoped delivery
Hand an agent the context for one stage of the work, within a budget, instead of a search result it has to trust.
Human review, recorded
Every stage ends at a human check, and the approval is tied to the output that was actually reviewed.
Memory as evidence
Your MemoryRelay memories feed in as evidence with their provenance — never silently promoted to instructions.
Coming soon
ICM workspaces are not generally available yet. Tell us what you would use them for at hello@memoryrelay.ai and we will reach out when early access opens.
In the meantime, the memory layer is available today — and what you store there is what your workspaces will draw on as evidence.
ICM is described in Interpretable Context Methodology by Jake Van Clief and David McDermott. MemoryRelay follows the ICM Architect reference (MIT licence). MemoryRelay's service is an independent implementation and is not endorsed by the authors.