Latest

LangChain's harness-engineering recipe treats evals as the training data for agents — but what do you do when you don't have evals, can't touch production customer data, and need to probe very specific corners of agent behaviour? You build a harness whose output is the dataset itself. This post walks through a synthetic-data-generation harness that runs Claude Code in a loop over Phoenix traces and generates edge-case companies designed to break the agent under test, grounded in what Meta-Harness and the LangChain canon actually say (and don't say) about how such loops should be built.

A source-level audit of Nous Research's hermes-agent. The 'self-improving' label cashes out as four orthogonal mechanisms — skill auto-generation, persistent memory, offline RL fine-tuning, and a pull-based update protocol — none of which is autonomous code rewriting. The interesting engineering is in the operational plumbing that makes all four robust in messy real-world deployments.

Building an autonomous alpha research system that uses Claude Code to generate factor hypotheses, evolve strategies through mutation and crossover, and subject everything to a gauntlet of statistical tests designed to catch false discoveries.

How Recursive Self-Improvement turns LLMs from static prediction machines into trading systems that get better on their own -- learning from news, research reports, and price data through evolutionary search, self-critique, and memory-based refinement loops.

How I built an evolutionary factor mining system that uses Claude Code to autonomously discover, backtest, and evolve commodity futures trading factors — and what it found.