Show HN: Shard – Stop watching one AI agent code for 45 min. Run four at once
https://github.com/nihalgunu/Shard
The biggest bottleneck in AI-assisted coding right now isn't model quality. It's that you give Claude Code or Cursor a complex task and then sit there watching it work for 30-60 minutes. You can't intervene, you can't do anything useful, you just wait.Shard takes that one big prompt and automatically decomposes it into a DAG of parallel sub-tasks. Each task gets exclusive file ownership so there are zero merge conflicts by design. Then it dispatches multiple agents across git worktrees simultaneously, merges in topological order, and self-heals test failures.
A 45 minute serial task becomes 4 agents running for 12 minutes.
Open source, cross-platform, agent-agnostic (Claude Code, Aider, Cursor).
(
2 comments
| on HN
)
one thing worth thinking through as shard matures: the management layer for those parallel agents running in worktrees. decomposition and dispatch is one side; the human staying oriented across 4 concurrent sessions - seeing what each is doing, switching between them, reviewing diffs inline, handling the merge step without context switching overhead - is the other.
i built pane for that layer specifically. fully open source (agpl-3.0), works with any cli agent. curious how you're handling the merge step when tasks in the dag have soft dependencies - like when two tasks touch overlapping files but aren't explicitly linked: github.com/Dcouple-Inc/Pane