An agentic system that reads research papers (local PDFs or arXiv), generates implementation plans, writes code, and reviews it — all orchestrated by a central Director agent built on the Claude Agent SDK.
15 files for Claude Code: implement, plan, read-papers, research and 11 more .
.claude/skills/architecture-design/SKILL.md A 25 tok .claude/skills/dependency-analysis/SKILL.md A 28 tok .claude/skills/extract-equations/SKILL.md A 28 tok .claude/skills/iterate-feedback/SKILL.md A 38 tok .claude/skills/paper-alignment-check/SKILL.md A 30 tok .claude/skills/pipeline-status/SKILL.md A 40 tok .claude/skills/sandbox-execute/SKILL.md A 28 tok .claude/skills/security-audit/SKILL.md A 32 tok .claude/skills/summarize-paper/SKILL.md A 35 tok .claude/skills/write-tests/SKILL.md A 24 tok .claude/commands/implement.md A 6 tok .claude/commands/plan.md A 7 tok .claude/commands/read-papers.md A 12 tok .claude/commands/research.md A 17 tok .claude/commands/review.md A 8 tok These files are Cenrax/researchswarm's own configuration — they tell Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.