monospace.studio
← Back to Product Labs

Active system · 2026

shulkerbox

A universal resource repository for product-development tools, skills, agents, hooks, scripts, and plugins.

Shulkerbox is the shared shelf for the agentic product-development system. It keeps reusable workflows, agent personas, subagents, hooks, scripts, templates, configs, and reference material in one tool-agnostic repo so they can travel across Claude Code, Copilot CLI, Gemini CLI, IDEs, and future development environments.

What problem does this solve?

AI-development workflows get trapped inside individual tools, local folders, and half-remembered prompts. Shulkerbox solves the drift by making reusable skills, agents, hooks, scripts, and templates portable.

What it can do

  • Centralise reusable AI-development resources instead of scattering them across individual tools or projects.
  • Link skills and hooks into local environments with symlinks, so updates propagate without copying files around.
  • Package repeatable workflows as skills: wrap-up, pickup, note-that, activity-log, weekly-review, and work-monitor.
  • Separate main-context skills, opinionated agent personas, and isolated subagents so different kinds of work get the right amount of context.
  • Run a personal productivity loop where notes, friction logs, activity logs, and weekly reviews surface which new skills should be built or pruned.
  • Provide starter templates for project conventions, notes, rules, scripts, and shared configuration.

Activity signals

  • Local source includes `skills/`, `agents/`, `.claude/agents/`, `hooks/`, `scripts/`, `plugins/`, `templates/`, `configs/`, `library/`, and `docs/`.
  • Architecture docs define the distinction between skills, agents, and subagents, including when each should be used.
  • Productivity-system docs connect activity logging, insight capture, weekly review, and skill backlog management into a self-improvement loop.
  • README positions the project as cross-tool infrastructure rather than a single IDE integration.

Why it matters

  • Shows the operating-system layer behind product experiments, not only finished UI.
  • Makes agent enablement portable across tools instead of trapped in one IDE.
  • Useful advisory evidence for teams trying to standardise AI-assisted delivery.

Recent commits

DateHashChange
2026-05-087fbac36Add tone of voice snippet
2026-04-212cbffe9Enforce branch workflow with pre-push hook
2026-04-21aa56555Add pre-commit verification, breaking change detection, and session resumption
2026-04-1423e0aa8Reorganize skills and agents into categorized structure
2026-04-135c52374Make wrap-up write to the activity log every session

Next move

Fold in the best reusable hooks, skills, and templates from the other active repos.