By ronsanzone
Deep-work pipeline: a 6-phase context engineering workflow that separates research from solutioning. Each phase runs in a fresh conversation for context isolation.
Produces a structured "how it works" writeup of existing code, with file:line citations on every claim and optional mermaid flow diagrams. Use when you need a durable technical artifact (data flow, control flow, integration points) — not a quick exploratory answer. Prefer over `Explore` when the result will be referenced or shared. Do not use for open-ended search.
Returns a categorized file-location inventory for a feature (implementation / tests / config / types / docs) — paths only, no analysis of contents. Prefer over `Explore` when you're planning changes and need to know WHERE code lives, not what it does. Use whenever you'd otherwise run grep/glob/ls more than once.
Returns 3–4 ready-to-copy code snippets (10–30 lines each, with file:line refs and matching tests) showing how a pattern is implemented in this codebase. Use when you're about to write new code that should match existing conventions. Prefer over `Explore` when you need concrete template examples to model after, not narrative answers.
Use when starting a new deep-work task. Decomposes a task description into objective research questions that avoid premature solutioning. Phase 1 of the deep-work pipeline.
Use when you have research questions from deep-work Phase 1. Objectively investigates the codebase to answer pasted questions without access to the original task description.
Use when deep-work Phase 2 research is complete. Combines research findings with the original task to explore design options, evaluate tradeoffs, and make decisions interactively.
Use when deep-work Phase 3 design decisions are complete. Maps chosen decisions to concrete file changes organized into implementable phases.
Use when deep-work Phase 4 structure outline is complete. Expands the outline into a detailed implementation plan with exact file paths, code patterns, tests, and validation commands.
Uses power tools
Uses Bash, Write, or Edit 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.
Structured, multi-phase workflows for Claude Code that separate research from solutioning. Two pipelines available:
Both pipelines produce durable artifacts at ~/notes/context-engineering/<repo>/<slug>/ and dispatch fresh subagents per task with two-stage review (spec compliance → code quality).
git clone [email protected]:ronsanzone/context-engineering-workflows.git ~/code/context-engineering-workflows
cd ~/code/context-engineering-workflows
# Install everything
./install.sh
# Or install one pipeline
./install.sh rpi # RPI skills + shared agents
./install.sh dw # DW skills + shared agents
Skills are copied to ~/.claude/skills/ and agents to ~/.claude/agents/. To update, pull and re-run ./install.sh.
# Add the marketplace (one-time)
/plugin marketplace add ronsanzone/context-engineering-workflows
# Install one or both pipelines
/plugin install rpi@context-engineering-workflows
/plugin install dw@context-engineering-workflows
To update, run /plugin marketplace update context-engineering-workflows, then reinstall.
/rpi-research <slug> <research query> # Investigate the codebase
/rpi-plan <slug> # Interactive plan creation (4 gates)
/rpi-implement <slug> # Execute the plan with subagents
/dw-01-research-questions <slug> # Decompose task into research questions
/dw-02-research <slug> # Investigate (bias firewall — no prompt access)
/dw-03-design-discussion <slug> # Explore design options
/dw-04-outline <slug> # Map decisions to file changes
/dw-05-plan <slug> # Detailed implementation plan + adversarial review
/dw-06-implement <slug> # Execute the plan
Each DW phase runs in a fresh conversation for context isolation.
Both pipelines use these specialized subagents:
| Agent | Purpose |
|---|---|
codebase-analyzer | Traces data flow, explains how code works, produces file:line-cited writeups |
codebase-locator | Finds where code lives — returns categorized file inventories |
codebase-pattern-finder | Finds similar implementations to model after — returns ready-to-copy snippets |
skills/ # Source of truth — edit here
├── shared/setup.sh # Artifact directory resolution (co-located into each skill at install)
├── deep-work/ # Pipeline overview (docs-only)
├── dw-*/ # DW pipeline phases
└── rpi-*/ # RPI pipeline phases
agents/ # Source of truth — edit here
├── codebase-analyzer.md
├── codebase-locator.md
└── codebase-pattern-finder.md
plugins/ # Generated — do not edit
├── dw/ # Marketplace plugin for DW
└── rpi/ # Marketplace plugin for RPI
scripts/sync-plugins.sh # Regenerates plugins/ from source
install.sh # Raw install to ~/.claude/
Edit files in skills/ and agents/ only. The plugins/ directory is generated by scripts/sync-plugins.sh and auto-staged by a pre-commit hook.
After cloning, configure the hook:
git config core.hooksPath .githooks
RPI pipeline: a streamlined 3-phase research-plan-implement workflow. Produces durable research artifacts and dw-05-format implementation plans.
npx claudepluginhub ronsanzone/context-engineering-workflows --plugin dwClaude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.
AI-powered development tools for code review, research, design, and workflow automation.
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows