Documents solved problems into team knowledge base with context, code examples, gotchas, and links. Use after workflows, incidents, or features to compound learnings.
npx claudepluginhub v1truv1us/ai-eng-system --plugin ai-eng-learningThis skill uses the workspace's default tool permissions.
Transform solved problems into reusable team knowledge. This skill ensures every problem your team solves makes future similar problems easier to solve. Used in the **Compound** phase of the spec-driven workflow (Phase 6) and can be called independently after completing any project or solving any significant issue.
Creates isolated Git worktrees for feature branches with prioritized directory selection, gitignore safety checks, auto project setup for Node/Python/Rust/Go, and baseline verification.
Executes implementation plans in current session by dispatching fresh subagents per independent task, with two-stage reviews: spec compliance then code quality.
Dispatches parallel agents to independently tackle 2+ tasks like separate test failures or subsystems without shared state or dependencies.
Transform solved problems into reusable team knowledge. This skill ensures every problem your team solves makes future similar problems easier to solve. Used in the Compound phase of the spec-driven workflow (Phase 6) and can be called independently after completing any project or solving any significant issue.
Compounding Engineering: Each unit of work should make future work easier. When you solve a problem, document it. When the next person encounters a similar problem, they solve it 10x faster. Over time, your knowledge base becomes your competitive advantage.
/ai-eng/review)Without systematic knowledge capture:
With this skill:
Before writing, collect:
The Problem (what was hard?)
The Solution (what did we do?)
Evidence (what's the proof?)
Gotchas (what could trip people up?)
Reach (who needs to know?)
File structure:
docs/solutions/
├── category/
│ ├── README.md (category index)
│ └── [topic].md
├── performance/
│ ├── redis-query-optimization.md
│ ├── database-indexing-strategy.md
│ └── bundle-size-reduction.md
├── security/
│ ├── csrf-protection-in-forms.md
│ └── secrets-management.md
├── deployment/
│ ├── zero-downtime-migrations.md
│ └── canary-releases.md
└── debugging/
├── memory-leak-investigation.md
└── race-condition-detection.md
Minimum content per solution:
# [Problem Title]
**Date Added**: 2026-02-08
**Category**: [performance|security|deployment|debugging|architecture]
**Audience**: [frontend|backend|devops|all]
**Complexity**: [easy|medium|hard]
**Time to Implement**: [estimate, e.g., "2 hours"]
## Problem
[What was the issue? Why was it hard? What made it non-obvious?]
### Example
[Show the problem in action - code, screenshot, error message]
## Solution
[What did we do to fix it? Why does this work?]
### Implementation
[Step-by-step walkthrough or code example]
### Why This Works
[Explain the underlying principle]
## Gotchas
- **Gotcha 1**: What could trip people up?
- How to avoid: [specific advice]
- **Gotcha 2**: Common mistake
- How to avoid: [specific advice]
## Alternatives Considered
- **Approach A**: Why we didn't use it
- **Approach B**: Why we didn't use it
- **Our Choice (C)**: Why this one is best
## Verification
How to know if you've solved it correctly:
- [ ] Test case passes
- [ ] Performance metric achieved
- [ ] No regressions in related areas
- [ ] Code review approval
## Related Solutions
- [Sibling solution](./related.md)
- [Related pattern](./pattern.md)
- External: [Link to article/docs](url)
## Questions for Future Readers
If you're implementing this, ask:
- Do you have [X setup]?
- Are you on [specific version]?
- Is [config] set to [value]?
**Confidence in this solution**: 0.9/1.0
**Missing/Uncertain**: Minor edge cases for version < 2.0
**Last Updated**: 2026-02-08
**Updated By**: [name]
After creating a solution, update:
Category index (docs/solutions/[category]/README.md):
## [Category] Solutions
- [Problem Title](./problem-title.md) - Brief description
- [Another Problem](./another.md) - Brief description
Global index (docs/solutions/README.md):
## All Solutions
### [Category]
- [Solution](./category/solution.md)
Git commit with clear message:
git commit -m "docs: add solution for [problem] in [category]"
/ai-eng/review/ai-eng/compound "database query optimization breakthrough"
# In chat after incident resolution
Use knowledge-capture skill to document: API timeout under load and fix
# Team retrospective - capture learnings
Capture: What we learned about microservices deployment
# In OpenCode, call knowledge-capture when appropriate
capture_knowledge(
problem="N+1 queries in user profile endpoint",
solution="Implemented DataLoader pattern",
category="performance"
)
Before publishing a solution document, verify:
# Redis Connection Pooling for High Throughput
**Problem**: Our API was exhausting Redis connections under 1k concurrent users, causing timeouts.
**Solution**: Implemented connection pooling with these specific settings: [settings] and switched from individual connections to pool pattern.
**Code Example**:
[Real code that worked]
**Gotchas**:
- Setting max_idle too low caused connection churn
- Connection timeout needs to be 2x higher than request timeout
- Must enable keep-alive or connections drop after 5 minutes
**Confidence**: 0.95 - Tested with load up to 5k concurrent
**Uncertainty**: Behavior under 10k+ concurrent unknown, might need sharding
# Performance Improvements
We made our system faster by optimizing things.
Confidence: ???
After 6 months of using this skill:
Track:
❌ Too Broad: "How to debug" → ✅ "Race condition detection in Goroutines"
❌ Too Deep: 50 paragraphs → ✅ 5-10 paragraphs + code example
❌ No Context: "Do this" → ✅ "Do this because X, which solves Y"
❌ Untested: "Should work" → ✅ "Tested in production with Z"
❌ No Updates: Created once, never touched → ✅ Updated yearly
| Metric | Target | Value |
|---|---|---|
| Solutions documented | 50+ | [Check] |
| Average confidence | 0.85+ | [Check] |
| Reuse rate | 30%+ | [Check] |
| Onboarding time | -30% | [Check] |
| Time to solve known problems | -50% | [Check] |
/ai-eng/compoundIn Claude Code, the /ai-eng/compound command uses this skill automatically:
/ai-eng/compound "redis connection pooling breakthrough"
# Automatically:
# 1. Gathers context about the solution
# 2. Creates docs/solutions/[category]/[topic].md
# 3. Updates category index
# 4. Updates global index
# 5. Creates git commit
After completing any meaningful work:
Compounding engineering isn't about working harder—it's about working smarter by capturing and reusing knowledge.