Description
How useful an AI coding assistant is comes half from the model and half from the workflow it operates in. Left unguided, it makes the same set of mistakes: unstructured prompts, overflowing context, tool sprawl, no error handling, multi-agent systems for single-agent problems. That is not the model's fault — it is a workflow without discipline.
Maestro Workflow is a skill pack that supplies that discipline. Installed, it adds a set of slash commands across ten-odd AI clients — Claude Code, Cursor, Copilot, Gemini CLI and more:
What sets it apart from a pile of loose prompts is memory and an audit trail: decisions, audit history and session context persist across sessions rather than being forgotten when the window closes. It also ships curated anti-patterns that tell the AI explicitly what not to do, not just what to do, and every command recommends a next step so there are no dead ends.
Install with one line,
25 workflow commands: categorised slash commands for diagnosis, evaluation, refinement, fortification and reflection, each aimed at common failures like messy prompts, context overflow and missing error handling.
Core skill plus references: one agent-workflow core skill and seven domain reference files that encode what a reliable workflow should look like as reusable guidance.
Persistent memory layer: decisions, audit trail and session history survive across sessions, so the AI remembers what was decided and changed without re-briefing.
Anti-pattern catalogue: curated anti-patterns tell the AI which approaches to avoid, cutting down predictable, repeated mistakes.
Context-gathering protocol: a .maestro.md or .maestro/context.md file gives every command project-specific awareness rather than generic answers.
Next-step suggestions: each command recommends a follow-up, chaining loose commands into a coherent flow with no dead ends.
Many clients, several entry points: compatible with Cursor, Claude Code, Copilot, Gemini CLI and ten clients in all, via npx install, a VS Code extension and an MCP server.
Maestro Workflow is a skill pack that supplies that discipline. Installed, it adds a set of slash commands across ten-odd AI clients — Claude Code, Cursor, Copilot, Gemini CLI and more:
/diagnose finds workflow issues, /streamline removes unnecessary complexity, /fortify adds error handling, /refine does a final quality pass. Twenty-five commands span diagnosis, evaluation, refinement, fortification and reflection, backed by one core skill and seven domain reference files.What sets it apart from a pile of loose prompts is memory and an audit trail: decisions, audit history and session context persist across sessions rather than being forgotten when the window closes. It also ships curated anti-patterns that tell the AI explicitly what not to do, not just what to do, and every command recommends a next step so there are no dead ends.
Install with one line,
npx skills add sharpdeveye/maestro, with a VS Code extension and an MCP server as alternative ways in. MIT licensed.Features
25 workflow commands: categorised slash commands for diagnosis, evaluation, refinement, fortification and reflection, each aimed at common failures like messy prompts, context overflow and missing error handling.
Core skill plus references: one agent-workflow core skill and seven domain reference files that encode what a reliable workflow should look like as reusable guidance.
Persistent memory layer: decisions, audit trail and session history survive across sessions, so the AI remembers what was decided and changed without re-briefing.
Anti-pattern catalogue: curated anti-patterns tell the AI which approaches to avoid, cutting down predictable, repeated mistakes.
Context-gathering protocol: a .maestro.md or .maestro/context.md file gives every command project-specific awareness rather than generic answers.
Next-step suggestions: each command recommends a follow-up, chaining loose commands into a coherent flow with no dead ends.
Many clients, several entry points: compatible with Cursor, Claude Code, Copilot, Gemini CLI and ten clients in all, via npx install, a VS Code extension and an MCP server.
