This is a workflow discipline layer for Claude that enforces pragmatic engineering practices when working on code projects. It pushes sequential task execution, terse communication, and safety rails like requiring confirmation before deleting databases or pushing to production. The guidelines discourage blame-oriented debugging and encourage root cause analysis when you're circling the same problem. Notable specifics include testing after every task, analyzing database schemas before queries, and reusing scripts instead of rewriting them. It's opinionated about response style (no emoji, no "production ready" claims) and treats AI assistance as a workhorse rather than a conversational partner. Useful if you want Claude to act more like a cautious senior engineer and less like an enthusiastic assistant.
npx -y skills add microck/ordinary-claude-skills --skill "Claudeisms Operational Guidelines" --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills