joslat 

The comprehensive .NET evaluation toolkit for AI agents—tool usage validation, RAG quality metrics, stochastic evaluation, model comparison, and red team security testing. Built for Microsoft Agent Framework... More information

MAF Doctor — a .NET global tool and MCP server for Microsoft Agent Framework (MAF). Diagnoses, explains, and auto-fixes MAF agents and workflows: A-F health grades, deterministic Roslyn rewriters, anti-pattern... More information

Abstractions and domain models for AgentMemory for .NET, a Neo4j-backed agent-memory library.

Roslyn analyzers for Microsoft Agent Framework (MAF) — write-time enforcement of fan-out safety, secure credentials, and observability defaults (MAF001-MAF003). Companion to the maf-doctor .NET global tool /... More information

Framework-agnostic core for AgentMemory for .NET — memory orchestration, context assembly, extraction-pipeline coordination, and entity resolution (no Neo4j driver dependency).

Neo4j persistence and retrieval for AgentMemory for .NET — repositories, schema bootstrapper, and GraphRAG retrievers (vector, fulltext, hybrid, graph).

LLM-based structured extraction backend for AgentMemory for .NET — extracts entities, relationships, facts, and preferences from conversations.

Entity enrichment for AgentMemory for .NET — Nominatim geocoding and Wikimedia/Diffbot enrichment with caching, queuing, and rate limiting.

AgentEval CLI — terminal + CI/CD entry point for the .NET evaluation toolkit. The agenteval init command bootstraps a .agenteval/ workspace; agenteval bench... More information

MAF autopilot MCP server — Tools + Resources + Prompts + Sampling for MAF 1.3.0 migrations. Answers "is this API safe?", exposes migration guide as resources, and uses sampling for multi-step agentic analysis.

Azure AI Language extraction backend for AgentMemory for .NET. Entity recognition uses Azure NER; fact, preference, and relationship extraction are heuristic (Azure Text Analytics has no dedicated API for... More information

Convenience meta-package for AgentMemory for .NET, a Neo4j-backed agent-memory library for AI agents. References all essential assemblies so consumers need only a single package reference.

OpenTelemetry-based observability decorators for AgentMemory for .NET. Opt-in tracing, metrics, and structured logging.

Microsoft Agent Framework (MAF) adapter for AgentMemory for .NET — AIContext provider, chat message store, and memory tools.

Semantic Kernel adapter for AgentMemory for .NET, a Neo4j-backed agent-memory library.

MCP server for AgentMemory for .NET — exposes agent-memory tools to MCP clients via the Model Context Protocol.

Optional Neo4j Graph Data Science (GDS) analytics for AgentMemory for .NET — PageRank (memory importance) and Louvain community detection (topic clustering) over the entity graph. Opt-in; degrades to a graceful... More information

Roslyn analyzers for Microsoft Agent Framework (MAF) 1.3.0 — write-time enforcement of fan-out safety, secure credentials, and observability defaults. Pairs with the maf-autopilot MCP server.

Stage-1 NAMS (Neo4j Agent Memory as a Service) adapter for the Microsoft Agent Framework (MAF) -- an additive AIContext provider backed by the NAMS backend, kept in its own package (ADR-9) so neither... More information

MCP server tools for the NAMS backend of AgentMemory for .NET -- opt-in, capability-aware, never part of the automatic recall/persistence lifecycle.