ScientiaCapital/unsloth-mcp-server: how it configures its agents

ScientiaCapital/unsloth-mcp-server

MCP server for LLM fine-tuning with Unsloth. 33 tools, 180 tests, RunPod GPU integration. Fine-tune 2x faster with 80% less memory.

23 files for Claude Code: observer-full, observer-lite, begin, build and 19 more — 378 tokens loaded in every session.

Instructions

unsloth-mcp-server CLAUDE.md CLAUDE.md A 378 tok

Skills

adaptive-workflows .claude/skills/adaptive-workflows/SKILL.md A 51 tok
dataset-engineering .claude/skills/dataset-engineering/SKILL.md A 48 tok
model-deployment .claude/skills/model-deployment/SKILL.md B 52 tok
superbpe .claude/skills/superbpe/SKILL.md A 58 tok
training-optimization .claude/skills/training-optimization/SKILL.md A 45 tok
unsloth-finetuning .claude/skills/unsloth-finetuning/SKILL.md A 66 tok
unsloth-mcp-server .claude/skills/unsloth-mcp-server/SKILL.md C 43 tok
unsloth-tokenizer .claude/skills/unsloth-tokenizer/SKILL.md A 53 tok

Commands

begin .claude/commands/begin.md A 15 tok
build .claude/commands/build.md A 14 tok
done .claude/commands/done.md A 11 tok
end .claude/commands/end.md A 20 tok
fix .claude/commands/fix.md A 14 tok
go .claude/commands/go.md A 16 tok
interrupt .claude/commands/interrupt.md A 11 tok
pr .claude/commands/pr.md A 14 tok
review .claude/commands/review.md A 6 tok
ship .claude/commands/ship.md A 9 tok
simplify .claude/commands/simplify.md A 11 tok
status .claude/commands/status.md A 14 tok

Agents

observer-full .claude/agents/observer-full.md A 39 tok
observer-lite .claude/agents/observer-lite.md A 51 tok

These files are ScientiaCapital/unsloth-mcp-server'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.