WHAT SHIPS TODAY
Every capability on this page ships today, with a measured answer behind it — a memory with rules, sweeps that compound, and a door any MCP-capable agent can open.
DEEP FIX
The daily cycle asks what work arrived. A sweep asks the other half: what is wrong with what we already built, and does it keep being wrong the same way? Auditors run in waves, and every wave inherits the finding types the waves before it discovered — so the rubric is a permanent asset of the product, not a linter's fixed opinion.
Findings close as fixed, refuted on inspection, owner decision, or upstream — a fixer who correctly rejects a finding is a first-class outcome, not a failure.
ONE MEMORY PER TEAM
A workspace binds to a repository with a committed file, so every member's agent — on any machine — opens the same memory. Teammates who only need to see don't need a member seat.
Viewer seats, server-enforced Read-only, enforced on the server rather than hidden in the UI — ten viewer seats per workspace by default, adjustable per workspace.
Share a board by link A daily run can be shared read-only with someone outside the org entirely — no account required. Links expire, revoke without breaking a live run, and every view is recorded.
Machines on record Every machine that ever synced is on record — a wiped laptop's history is never erased, just joined by its successor.
MEMORY WITH A BUDGET
A memory system fails by flooding. SF Recall refuses the flood: corrections supersede rather than stack, and standing decisions draw from a hard budget, so importance keeps meaning something.
Supersede, never stack A near-duplicate deposit is refused with the existing answer attached — you replace it or you drop yours.
A standing-decision budget When the budget is full, something must be demoted before anything new is promoted. No infinite pile of 'critical'.
Attention items are promises Unfinished business reappears at every standup until someone closes it — it cannot quietly rot in a backlog.
ANY AGENT, ANY MODEL
The memory mounts over MCP with OAuth 2.0, PKCE and dynamic client registration — an agent registers itself and gets the whole surface. There is no per-vendor integration to wait for, because the protocol is the integration.
Deposit, recall, supersede, standup The primitives are generic. Nothing in them assumes you are writing code — the daily cycle and the deep-fix rubric are built ON this memory, not baked into it.
A REST surface as well as MCP An application that is not an MCP client can still deposit and recall. If you are wiring model calls into your own software, you can give that a memory rather than build one.
One live answer per question Corrections supersede rather than stack, and every memory points back to the session that wrote it — so a second agent inherits a settled answer, not a pile of contradictions.
Provenance on every memory Each memory points back to the session that wrote it, so 'why do we do it this way?' has a checkable answer.
Claude Code captures conversations today Automatic transcript capture is written as a pluggable adapter and Claude Code is the first. Every other capability — standup, deposit, recall, board, sweeps — already works from any MCP-capable agent.
