Build software better, together

JustVugg / agentmw

Open-source middleware for AI agents — catches mid-run failures,compresses stale context, and grows a reasoning library across runs. Any model, any framework.

python mcp openai developer-tools reasoning ai-agents llm prompt-engineering anthropic openrouter ollama llm-observability agent-monitoring agent-orchestration mcpserver agent-middleware

Updated May 19, 2026
Python

Zoria-Lind / dsh-token-optimizer

Star

Layered token-optimization pipeline for DeepSeek Harness: output ladder, MCP lazy loading,compaction driver, cache-hit reporting. Built on real DSH plugin APIs; ~40-60% input saved in long sessions.

plugin agent tokens awesome-list token cost-control dsh cost-optimization token-usage deepseek context-management agent-memory token-optimization cost-control-context-hints agent-middleware deepseek-harness dsh-plugin deepseek-harness-plugin

Updated Oct 4, 2026
JavaScript

Zoria-Lind / dsh-behavior-enhancer

Star

Behavior-management plugin for DeepSeek Harness: tool-call discipline prompt section, failure-triggered parallelism convergence (pool drops to 1, auto-restores), consecutive-failure user intervention. Complements dsh-token-optimizer; works standalone.

agent awesome awesome-list agents dsh agent-behaviors deepseek agent-behavior agent-skills agent-skill agent-behavior-development agent-middleware agent-behavior-rule-builder deepseek-harness dsh-plugin deepseek-harness-plugin deepseek-harness-plugins

Updated Oct 19, 2026
JavaScript

oSEAItic / harbor

Star

Open-source agent middleware — shared memory, credential isolation, and schema learning for AI agents. What one agent learns, every agent knows.

open-source mcp gemini cursor shared-memory codex ai-agents model-context-protocol mcp-server agent-native agent-memory context-protocol claude-code context-engineering schema-learning credential-isolation agent-middleware

Updated Aug 6, 2026
Go

agentmiddleware / agent-middleware

Star

Permits and signed receipts for agent tool calls. One dispatch, one charge, verifiable offline.

mcp ai-agents idempotency model-context-protocol agent-security agent-middleware signed-receipts

Updated Oct 26, 2026
Python

whiterabb17 / mentalist

Star

Mentalist: (DeepAgent in rust) High-performance Rust agentic middleware & execution environment for autonomous AI agents. Implements DeepAgent methodology with execution hooks, explicit planning, sandboxed tool execution, and full session serialization. Built for production-ready agent systems.

rust sandboxing memory-management autonomous-agents ai-agents agent-framework deepagent tool-execution agent-middleware state-serialization

Updated Apr 16, 2026
Rust

agentmiddleware / amw-demo

Star

Public API demo of Agent Middleware: permit, governed call, safe retry, offline receipt check.

mcp ai-agents idempotency model-context-protocol agent-security agent-middleware signed-receipts

Updated Oct 24, 2026
Python

agentmiddleware / agent-middleware-api

Star

MCP authorization gateway for agent-to-tool actions: scoped permits, at-most-one gateway dispatch and debit, signed receipts, and audit evidence.

Updated Oct 24, 2026
Python

agentmiddleware / amw-paypal-agent-demo

Star

PayPal sandbox checkout agent governed by Agent Middleware: at most one capture, verifiable receipts.

mcp ai-agents idempotency model-context-protocol agent-security agent-middleware signed-receipts

Updated Oct 26, 2026
Python

ashutoshrana / confidence-escalation

Star

Framework-agnostic confidence-gated escalation middleware for LLM agents: multi-signal scoring (logprob, verbalized, tool risk), threshold policies, and escalation handlers for LangChain, CrewAI, AutoGen, and Google ADK.