Awesome list for AI agent harness engineering: tools, patterns, evals, memory, MCP, permissions, observability, and orchestration.
Python agent-harnessagent-memoryagent-orchestrationai-agent-harnessai-agentsawesome-listcontext-engineeringharness-engineeringmcp
Commits on main
Add Claw-Eval: rigorous community-verified eval harness with 300 human-verified tasks and Pass^3 methodology
eba1b02 ai-boost committed 6 months ago
Add dirac: surgical context curation harness that cuts API costs 50–80%
35e1384 ai-boost committed 6 months ago
Add agentmemory: cross-agent persistent memory layer for coding agents
e775801 ai-boost committed 6 months ago
Add vurb.ts: TypeScript framework for building production MCP servers with Presenter perception layer and FSM gating
009f6fe ai-boost committed 6 months ago
Add Mirage: unified virtual filesystem that exposes S3, Slack, Gmail and more as a single bash filesystem for agents
aa9d3d6 ai-boost committed 6 months ago
Add mcp-agent: production-grade framework for building agents with MCP
47010a3 ai-boost committed 6 months ago
Add deepclaude: ports Claude Code's agent loop to DeepSeek and other backends
d404f24 ai-boost committed 6 months ago
Add Future AGI: open-source unified platform for tracing, evals, simulations, guardrails, and gateway
e797920 ai-boost committed 6 months ago
Add Hive: YC-backed production harness for multi-agent execution DAGs with state persistence and crash recovery
32d8f49 ai-boost committed 6 months ago
Add deepsec: Vercel Labs' production-grade security harness for agentic vulnerability scanning with idempotent execution and SKILL.md context injection
842a81f ai-boost committed 6 months ago
Add Harmonist: drop-in multi-agent framework with mechanical protocol enforcement via IDE hooks
f9f60f6 ai-boost committed 6 months ago
Add browser-harness: self-healing browser harness where the agent edits its own scaffold during execution
348efc1 ai-boost committed 6 months ago
Add MemPalace: local-first AI memory system with structured palace architecture and best-in-class LongMemEval benchmarks
31384de ai-boost committed 6 months ago
Add open-multi-agent: TypeScript-native auto-DAG orchestration with MCP, token budgets, and context compaction
a7cb183 ai-boost committed 6 months ago
Add codebase-memory-mcp: tree-sitter knowledge graph MCP server replacing file-read loops with sub-millisecond structured queries
d5c53b2 ai-boost committed 6 months ago
Add DESIGN.md: Google Labs' specification for describing visual identity systems to coding agents with machine-readable design tokens
51f4a8e ai-boost committed 6 months ago
Add engram: single-binary persistent memory for coding agents with SQLite + FTS5 and 18 MCP tools
b453d0c ai-boost committed 6 months ago
Add OpenViking: ByteDance's context database enabling hierarchical context delivery through a filesystem paradigm
d707946 ai-boost committed 6 months ago
Add Trellis: progressive spec system replacing monolithic system prompts with on-demand standards across coding agent platforms
35ce02a ai-boost committed 6 months ago
Add OpenAI Symphony: orchestration layer for managing agent work with isolated per-issue workspaces and proof-of-work handoffs
39a3822 ai-boost committed 6 months ago
Add rasbt/mini-coding-agent: a minimal pure-Python coding agent harness
9d3253c ai-boost committed 6 months ago
Add CubeSandbox: Tencent Cloud's production-validated microVM sandbox for AI agents
673b8bb ai-boost committed 6 months ago
Round 20: add Token Savior MCP server for structural code navigation
4ea4f68 ai-boost committed 6 months ago
Round 19: add context-mode MCP server for tool-output sandboxing
3195d39 ai-boost committed 6 months ago
Round 17: add claude-devtools for agent debugging visibility
c405e6c Claude Code committed 7 months ago
Round 16: add agentic-stack cross-tool harness portability resource
0729620 Agent committed 7 months ago