Orchestration
LongHorizon-Harness
AMAP-ML/LongHorizon-Harness
LongHorizon-Harness is a loop-engineering layer that turns existing coding and computer-use agents into long-running workers, coordinating manager, executor, and auditor roles across verified rounds.
Topics
Related projects
loushang2kA method-native AI coding workbench and CLI with multi-model routing, persistent sessions, and verifiable delivery.
penguin-harness3.2kA recursive self-improvement framework where agents iteratively evolve prompts, skills, and harnesses against built-in benchmarks.
Codeman839A self-hosted mission control for AI coding agents that runs and orchestrates multiple CLI agents (Claude Code, Codex, OpenCode and more) in persistent tmux sessions, with auto-respawn, phased orchestration, cron jobs and live monitoring from any device.
aeon791An autonomous orchestration framework that runs skills across nine agent harnesses unattended via GitHub Actions.
hcom655A multi-agent coordination layer that lets independent coding agents like Claude Code, Codex, and OpenCode message, monitor, spawn, and orchestrate each other across terminals and machines.
cc-haha15.3kA local-first desktop workbench for Claude Code with multi-session review, permissions, models and remote access.