nicholls

Crosschecks Claude's code claims using Dafny formal verification for provably correct Python/Go code, plus semi-formal reasoning for structured code analysis.

30 skills · 5 agents · 1 MCP server

Feedback loops for coding agents: Loop Cards, experiment arms, cross-model review, PR shepherding, autonomy and spend dials, and GitHub Actions automations. Ported from github/lever.

16 skills · 1 agent

Meta prompts that help you discover and install curated GitHub Copilot agents, instructions, prompts, and skills from the awesome-copilot repository.

4 skills · 1 agent

Write and check non-fiction prose against a measured reference of human writing: several candidate drafts under a voice file, a pick by measured profile, and a check of specifics against supplied material.

2 skills

Multi-agent PR review swarm and review-triage skills for automated, multi-lens code reviews.

2 skills

Generate structured performance reports on plugins, skills, and agents by analysing Claude Code session conversations.

1 skill

Autonomous agent scheduling for GitHub issues — scans, queues, and launches Claude Code sessions to fix bugs, implement features, and refine issue descriptions.