agents
las7/memharness
Bi-temporal agent long-term memory: SQLite-backed MCP server with recall ranking, hybrid vector+FTS recall, and a source-staleness signal
paperSelf-rewarding agents that retrace failures
Agents that attribute their own errors and retrace to repair multi-step reasoning.
articleA field guide to AI agents in 2026
What 'agent' actually means now, and the patterns that work.
modelAgentCore-8B
A model post-trained for tool use and multi-step planning.
paperLong-horizon credit assignment in RL
A method for propagating reward across very long action sequences.
tutorialMake an agent that uses tools
Wire a model to a calculator and a search tool, safely.
repoagent-tools
A library of well-typed tools for LLM agents.
toolAgentTrace
A debugger that visualizes an agent's step-by-step reasoning.
newsOpen-source agent framework crosses 50k stars
The toolkit for building tool-using agents sees rapid community adoption.
reporetrace-agents
Reference implementation of self-retracing agents.
paperTool use without fine-tuning
Teaching models to call tools purely through in-context demonstrations.