By Garabed96
Personal development toolkit built on context economy — brainstorm, plan, execute, ship, QA. Every skill optimizes for minimal token spend with maximum signal.
Code quality review agent. Reviews git diffs for correctness, architecture, testing, and production readiness. Dispatched by context:execute for [MED] and [HIGH] tasks after spec compliance passes.
Fresh-context implementation agent. Given a task description and codebase context, implements the change, writes tests, and commits. Reports status back to the coordinator.
Code review agent. Reviews diffs for spec compliance and code quality. Dispatched for [MED] tasks (quality only) and [HIGH] tasks (spec + quality).
Critical thinking coach for architectural decisions and engineering growth. Pushes the user to reason about tradeoffs, identify hidden assumptions, map second-order effects, and articulate decisions at an architect level. Use when the user says "think through tradeoffs", "help me reason about this", "systems thinking", "challenge my thinking", "why this over that", "what's the tradeoff", "what are the tradeoffs", "am I overthinking this", "review my reasoning", or invokes /ctx-architect-growth.
Use when ambiguity is high, scope spans multiple systems, or you need blind-spot catching that same-context self-review can't provide. Escalation target from /ctx-brainstorm.
Use before building any feature. Lean tier for small-to-medium tasks — self-review, no subagents. Auto-escalates to /ctx-brainstorm-ss when complexity signals detected (3+ subsystems, high ambiguity, new domain).
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
Discussion and exploration mode. No code changes. Think out loud together, discover what to build, challenge assumptions. Use when the user wants to explore before committing to action.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Uses Bash, Write, or Edit tools
Uses Bash, Write, or Edit tools
Alpha — I'm building this in public. It works for my workflow. It may break for yours. Issues welcome.
A context engineering plugin for Claude Code. Controls what the model sees at every phase of development — from first idea to merged PR.
You: "I need to add search to the dashboard"
/ctx-discuss → Think it through — what kind of search? What are the tradeoffs?
/ctx-engineering → Make sure Claude understands the intent before moving forward
/ctx-brainstorm → Spec with blind spots caught before you write code
/ctx-plan → Tagged implementation plan with complexity budgets
/ctx-worktree → Isolated git worktree, deps installed, ready to go
/ctx-execute → Subagents implement each task, reviewers verify
/ctx-ship → Preflight checks, verification gate, PR created
# End of day
/ctx-park → Context saved — decisions, learnings, next steps
# Next morning
/ctx-grab → Pick up exactly where you left off
No re-explaining. No lost context. No "wait, what were we doing?"
/plugin marketplace add Garabed96/ctx-plugin
/plugin install ctx@ctx-plugin
Platform: macOS (tmux, iTerm2, Terminal.app). Core skills are POSIX-compatible. Windows untested.
Before building anything:
/ctx-discuss — think before you build. No code changes, just conversation. Explore the problem, challenge assumptions, discover what to build/ctx-engineering — make sure Claude clearly understands what you're saying. Flags prompts that are too brittle or too vague before they waste a cycleEither can come first. Once the direction is clear, move to brainstorm.
Full feature workflow:
/ctx-brainstorm — surfaces blind spots, writes a spec, and self-reviews (or escalates to a subagent reviewer for complex features)/ctx-plan — takes the spec and produces tasks tagged [LOW], [MED], or [HIGH] with agent budgets/ctx-worktree — spins up an isolated git worktree with env files copied and deps installed/ctx-execute — walks the plan task by task, dispatching implementer and reviewer agents based on complexity/ctx-ship — preflight risk scan, verification gate, PR creation. Each phase is gated — I approve before it continuesQuick fix workflow:
/ctx-worktree — isolate the change/ctx-verify — prove it works before claiming it does/ctx-ship — PR itWhen I get stuck:
/ctx-debug — blocks me from jumping to fixes before investigating root cause/ctx-tdd — enforces Red-Green-Refactor. Catches me when I write the implementation firstEnd of session:
/ctx-park — saves a handoff: what happened, what I learned, what's next. Also captures non-obvious learnings as memories that accumulate into skill-level gotchas over time/ctx-grab — next session, picks up the handoff and re-aligns contextThe plugin separates judgment from execution:
ctx-execute for implementation and reviewScripts do the mechanical work at shell speed. Skills decide when and why to call them. The model never wastes tokens interpreting bash it could skip.
Invoked with /ctx:<skill-name>.
Ideation
| Skill | Purpose |
|---|---|
ctx-brainstorm | Lean brainstorming with self-review. Auto-escalates to ctx-brainstorm-ss when complexity warrants it. |
ctx-brainstorm-ss | Tier 2 — subagent spec reviewer with fresh context, scope decomposition, design-for-isolation. |
ctx-discuss | Think before you build. No code changes — discuss direction, then move to /ctx-plan when ready. |
ctx-architect-growth | Critical thinking coach. Pushes you to reason about tradeoffs and map second-order effects. |
Planning & Execution
| Skill | Purpose |
|---|---|
ctx-plan | Produces a tagged implementation plan from a brainstorm spec. |
ctx-execute | Dispatches subagents per task — [LOW] 1 agent, [MED] 2, [HIGH] full review sandwich. |
ctx-ship | Gated pipeline: preflight → verify → PR. Each phase requires your approval. |
ctx-parallel | Ad-hoc parallel dispatch for independent tasks. |
Quality
npx claudepluginhub garabed96/ctx-plugin --plugin ctxPersistent context for AI coding assistants
Core workflow for structured development: Research → Plan → Implement → Validate with thoughts/ management
Plan-Build-Run — Structured development workflow for Claude Code. Solves context rot through disciplined subagent delegation, structured planning, atomic execution, and goal-backward verification.
Conductor: Context-driven development for Claude Code - Measure twice, code once
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Production-grade engineering skills for AI coding agents — covering the full software development lifecycle from spec to ship.