Perstack – Containerized harness, 5 tests with full logs and API cost
Containerized agent harness with TOML config instead of code-wired frameworks.
Jeju is an experimental local-first agent harness for developers who want to define agent behavior in config, run it with strict boundaries, inspect every effect, and improve it with evaluation evidence.
Manifest-driven agents with eval feedback loops when most harnesses are prompt-only.
Developers building bounded AI agents with auditable workflows
LangGraph · CrewAI · AutoGen
Containerized agent harness with TOML config instead of code-wired frameworks.
Iteratively improves agent harnesses from 67% to 87% on tau-bench using production traces.
YAML contracts enforce agent behavior where Guardrails and LMQL focus on outputs.
tmux-based agent topology you can save and restore like a Docker compose file.
Named failure modes for AI agents, but it's markdown files—not an actual tool or implementation.
Parallel agent execution with 4.3x speedup and .nvmrc-style version pinning per project.