Skip to content

Conversation

@nhorton
Copy link
Contributor

@nhorton nhorton commented Jan 29, 2026

Summary

Adds a complete DeepWork implementation of the Compound Engineering methodology, inspired by Every Inc's Compound Engineering Plugin. This job provides a structured workflow for high-quality engineering work that compounds knowledge over time rather than accumulating technical debt.

Key Changes

  • New job: library/jobs/compound_engineering/ with complete implementation
    • job.yml: Job definition with 4 steps (plan, work, review, compound) and 3 workflow combinations
    • readme.md: Comprehensive documentation including:
      • Overview of the Compound Engineering philosophy
      • Installation and usage instructions
      • Detailed analysis of what DeepWork provides automatically vs. what the original plugin had
      • 10 feature gaps and 10 suggestions for improving DeepWork based on this implementation
      • Philosophy and credits
    • steps/plan.md: Create detailed implementation plans through idea refinement, codebase research, and prior learning checks
    • steps/work.md: Execute plans systematically with task tracking and incremental commits
    • steps/review.md: Multi-perspective code review with 8 specialized review angles
    • steps/compound.md: Document solutions and learnings for future reuse

Implementation Details

The Four-Step Workflow

  1. Plan (40%): Transform feature ideas into detailed implementation plans with codebase research
  2. Work (20%): Execute plans with task tracking and continuous testing
  3. Review (30%): Multi-perspective code review before merging
  4. Compound (10%): Document solutions and learnings for institutional knowledge

Core Philosophy

  • 80/20 split: 80% planning and review, 20% execution
  • Knowledge compounds: Each solved problem becomes searchable documentation for future work
  • Multiple perspectives: Reviews use 8 different specialized angles (security, performance, patterns, etc.)
  • Incremental execution: Small commits with continuous testing maintain quality

Notable Features

  • Interactive planning with codebase research using sub-agents
  • Multi-perspective code review with severity categorization (P1/P2/P3)
  • Structured solution documentation in docs/solutions/ with categories
  • Quality criteria for each step to ensure consistency
  • Comprehensive readme analyzing gaps and suggesting DeepWork improvements

Why This Matters

This job demonstrates how DeepWork can replicate sophisticated AI engineering workflows while being simpler and more portable than the original plugin. The readme's analysis of gaps and suggestions provides valuable feedback for improving the DeepWork platform itself.

The Compound Engineering methodology is particularly valuable for teams wanting to reduce technical debt over time through systematic quality practices and knowledge capture.

https://claude.ai/code/session_01EKw7Tc1vwYAH96kaDfq8xY

Adds a DeepWork implementation inspired by Every Inc's Compound Engineering
Plugin (https://github.com/EveryInc/compound-engineering-plugin).

The job implements the core methodology: "Each unit of engineering work
should make subsequent units easier—not harder."

Includes 4 steps:
- plan: Transform ideas into detailed implementation plans with research
- work: Execute plans with task tracking and incremental commits
- review: Multi-perspective code review before merging
- compound: Document solutions and learnings for future reuse

The README documents:
- What DeepWork does automatically (8 features not needing re-implementation)
- What the original plugin supports that this version does not (12 gaps)
- Suggestions for improving DeepWork based on the gaps (10 recommendations)

https://claude.ai/code/session_01EKw7Tc1vwYAH96kaDfq8xY
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants