memmesh-dotnet

Official .NET SDK for MemMesh — memory + prediction for AI agents. Semantic recall, a bi-temporal knowledge graph, belief revision, reflection, and calibrated forecasting.

dotnet add package MemMesh

Quickstart

using MemMesh;

var mm = new MemMeshClient("sk-...", "proj_...");

// Remember something — hand the engine the raw turn; it extracts what's worth
// keeping (filler comes back with an empty `Saved`).
var observed = await mm.Memory.ObserveAsync(
    text: "Sarah said she prefers email over phone for anything non-urgent.");
Console.WriteLine($"kept {observed.Saved.Count} of {observed.CandidateCount} candidates");

// Recall it, semantically
var hits = await mm.Memory.SearchAsync("how to reach sarah", limit: 5);
foreach (var h in hits) Console.WriteLine(h.Content);

// Synthesize higher-order insights, with provenance
var res = await mm.Memory.ReflectAsync(maxInsights: 3);
foreach (var i in res.Insights) Console.WriteLine($"{i.Content} ({i.Confidence:P0})");

// Point-in-time knowledge graph
var edges = await mm.Context.QueryGraphAsync(asOf: "2026-03-01T00:00:00Z");

Surface

mm.Memory (observe/create/search/list/update/delete/stats/confirm/promote/feedback, Reflect, PrefetchRelated, Dedup) · mm.Lattice (ExtractPatterns/MineMemories/ GetPattern/ListPatterns/GetContext/RunMonitorTick/GetMonitorStatus/Predict/PredictTarget/ GetProfile/GetCohort/PredictByCohort/Estimate/GetCalibration) · mm.Context (Build/BatchBuild/QueryGraph) · mm.Events · mm.Alerts · mm.Learning · mm.Typed · mm.Compliance · mm.Health.

Config: new MemMeshClient(apiKey, projectId, baseUrl?, httpClient?). API errors throw MemMeshException (.Status, .Body).

Apache-2.0 · memmesh.ai · docs