Happy slice of mcp pizzaMCP.pizza

ContextLattice

MCP.Pizza Chef: sheawinkler

ContextLattice keeps a durable record of what you are working on — the goal, the decisions already made, the risks, and the next step — stored on your own machine rather than in someone else's cloud. Open a new chat, or switch to a different assistant entirely, and it rebuilds that picture instead of making you re-explain everything. It also tracks which past approaches actually worked. Installing it means running a container stack yourself, so plan for a technical afternoon.

Coding
Notes

Use This MCP server To

Resume a project without re-explaining it from scratch Carry my working context between different AI assistants Look back at the decisions I made earlier in a project Keep my notes and history on my own computer Save a short progress note at the end of a session See which past approaches actually worked before trying again

README

ContextLattice

ContextLattice editorial website hero showing a live local context field

Open an agent. Already there.

The local-first intelligence layer that gives AI agents durable continuity, explainable retrieval, portable context, and verified learning across harnesses.

CLI first ContextLattice v4.0.8 Local-first runtime Apache License 2.0

Quickstart · How it works · Agent setup · Architecture · Docs · Updates

Stop replaying the brief

Models can reason. Harnesses can act. Neither reliably retains the mission when a chat, model, tool, account, or computer changes.

ContextLattice gives that work a durable, inspectable context layer. It reconstructs the active objective, selects the evidence that matters, carries it safely, and records what actually worked—without turning every prompt into a transcript dump or making cloud storage mandatory.

Capability What changes
Durable continuity Reopen the objective, decisions, repository state, risks, proof, and next move as one bounded packet.
Explainable retrieval Rank evidence by impact per token and expose source coverage, omissions, opposition, degradation, and receipts.
Portable context Move signed, least-privilege continuation across agents and machines while keeping execution and transport caller-owned.
Verified skill evolution Discover skills without loading every file, evaluate repeated wins on holdouts, and require review before promotion.
Privacy-bounded Aggregate Signal Learn from explicitly opted-in, clipped statistics while raw memory remains local; production activation stays hard-blocked pending independent privacy and utility review.

The CLI is the primary interface. The dashboard makes behavior and proof visible. HTTP and MCP are companion integration surfaces for applications and harnesses.

How it works

ContextLattice FAQ

Which assistants does this work with?
One setup command wires it into Codex, Claude Code, OpenCode, and several other coding agents at once. It does not install those agents for you.
Does my work leave my computer?
No. It is built local-first, your saved context stays on your machine, and cloud storage is never required.
Can I use this to stop repeating myself every session?
Yes — that is the whole point. It reopens the goal, the decisions, and the next move as one bundle when you start again.
How hard is setup?
This one is for the technically comfortable. You copy the project down, duplicate a settings file, and start a local container stack.
What do I need installed first?
macOS, Linux, or Windows through WSL2, plus a container runtime such as OrbStack or Docker Compose.
Can I see why it surfaced something?
Yes. A dashboard shows what was retrieved and how it ranked, and the command line reports where each piece came from and what was missing.