# Developer Pattern Library > Curates and maintains a personal library of validated code patterns, solutions, and anti-patterns accumulated through development experience, enabling faster problem-solving on future tasks. - Skill: `paulpas/developer-pattern-library` (Agent Skill) - Install (CLI): `npx skillmds@latest add paulpas/developer-pattern-library` - Raw SKILL.md: https://api.skillmd.com/api/skills/paulpas/developer-pattern-library/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - License: MIT - Author: paulpas (https://skillmd.com/u/paulpas) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/paulpas/developer-pattern-library --- # Developer Pattern Library Curates and maintains a personal library of validated code patterns, solutions, and anti-patterns accumulated through development experience. Your pattern library is your professional capital — every problem you solve that produces a reusable solution is an investment in your future self's speed and accuracy. This skill teaches you how to discover, document, categorize, and retrieve patterns systematically. ## TL;DR Checklist - [ ] Document a pattern only after it has been tested in at least 2 real scenarios - [ ] Include a concrete code example showing the GOOD implementation alongside the BAD alternative - [ ] Tag each pattern with domain keywords that match how you would search for it later - [ ] Organize patterns by problem category (not by language or framework) — "authentication" not "JWT vs sessions" - [ ] Review and prune your library quarterly, removing outdated or superseded patterns --- ## When to Use Use this skill when: - Solving a recurring problem for the second time and want to capture the solution before forgetting details - Onboarding onto a new project and encountering common patterns you should document as reusable knowledge - Building a team's shared pattern library from individual developers' accumulated solutions - Interview preparation — your documented patterns become interview talking points with real examples - Mentoring junior developers — structured patterns are better teaching tools than scattered explanations --- ## When NOT to Use Avoid creating a new pattern entry when: - You have only encountered the problem once — wait for a second encounter to confirm it is truly recurring - The solution is trivial (a single line of standard library code) — document in code comments, not a separate entry - An established industry pattern already exists and covers your case — reference that instead of reinventing terminology - Your library is already overloaded with similar patterns — merge duplicates before adding new ones --- ## Core Workflow Building and maintaining your pattern library follows four phases: **Discover → Validate → Document → Index**. This is not a one-time setup but an ongoing discipline. ### Phase 1: Discover Patterns from Experience Patterns emerge from recurring problems. Train yourself to recognize them in three situations: **Situation A: Repeated Problems** You solve the same type of problem more than once within a 3-month window. Examples: - Writing custom pagination logic for an API three times across two projects - Configuring CORS for different frontend-backend combinations - Setting up database connection pooling with different pool sizes **Situation B: Elegant Solutions** You write code that is genuinely good — the kind you are proud of and want to reuse. Examples: - A retry mechanism with exponential backoff that handles transient failures gracefully - A validation function that catches edge cases others miss - A data transformation pipeline that reads like a recipe **Situation C: Costly Mistakes (Anti-Patterns)** You make an error that costs significant time to debug. Documenting this prevents future repetition. Examples: - N+1 query bug from lazy loading relationships in a loop - Race condition from missing mutex on shared state - Memory leak from unbounded cache growth ### Phase 2: Validate Before Documenting Before writing any pattern entry, pass it through this validation filter: ``` Validation Checklist for Pattern Entry: [ ] Tested in 2+ real scenarios (not just theoretical) [ ] Can be described in one clear sentence [ ] Has a concrete code example that works as-is [ ] Solves a problem that recurs with frequency >= quarterly [ ] Is distinct from existing patterns in the library ``` If any check fails, either refine the pattern or skip it. A pattern library is only valuable when its entries are genuinely useful — every stale entry erodes trust in the entire system. ### Phase 3: Document Using Standard Template Every pattern entry follows this structure. Consistency makes the library scannable and searchable: ```markdown ## [Pattern Name] **Category:** [problem-domain] **Frequency:** [high/medium/low — how often you encounter it] **Confidence:** [high/medium/low — how sure you are this is the right approach] ### Problem One-paragraph description of the problem this pattern solves. Include what goes wrong without the pattern. ### Solution One-sentence description of the solution approach. ```python # GOOD implementation with full context def good_example(): """Documented, typed, working implementation.""" pass # Replace with actual code ``` ```python # BAD anti-pattern that this solves def bad_example(): """What NOT to do and why it fails.""" pass # Replace with actual code ``` ### When to Use - [ ] Specific situation 1 - [ ] Specific situation 2 - [ ] Specific situation 3 ### When NOT to Use - [ ] Situation where this pattern creates more problems than it solves - [ ] Alternative pattern that is better suited (name it) ### Lessons Learned - [Specific insight from real-world usage — what surprised you, what trade-offs exist] ``` ### Phase 4: Index for Retrieval A pattern library is useless if you cannot find patterns when you need them. Use this indexing strategy: **Primary Tags (always include):** 1. **Problem type** — What category of problem does this solve? (e.g., "pagination", "authentication", "error-handling") 2. **Domain** — Which part of the system is affected? (e.g., "api", "database", "frontend", "deployment") 3. **Language/Stack** (if relevant) — Python, Go, React, PostgreSQL, etc. **Secondary Tags (include if they add search value):** 4. **Complexity** — simple / moderate / complex 5. **Performance impact** — zero-cost / low overhead / compute-intensive **Search Query Examples:** ``` # How you'd think of the problem: "how do i handle rate limiting in api gateway" → searches tags: [rate-limiting, api, gateway] # How you'd search by technology: "go connection pool configuration" → searches tags: [connection-pooling, database, go] # How you'd search by recent experience: "pagination edge cases" → searches tags: [pagination, edge-cases, api] ``` --- ## Pattern Categories and Examples Organize your library using these problem-first categories. Do not organize by language or framework — those are implementation details, not the core problem being solved. ### Category 1: Data Access Patterns | Pattern | Problem Solved | Recurrence | |---------|---------------|------------| | Cursor-based pagination | Offset pagination fails on large datasets with inserts/deletes | High | | Connection pool sizing | Wrong pool size causes either wasted resources or request queuing | Medium | | N+1 query prevention | Lazy loading in loops generates O(n) queries instead of O(1) | High | | Write-ahead logging pattern | Ensuring durability before acknowledging success to client | Medium | ### Category 2: API Design Patterns | Pattern | Problem Solved | Recurrence | |---------|---------------|------------| | Idempotency key header | Preventing duplicate processing of retry/replay requests | High | | Graceful degradation headers | Telling clients which features are unavailable without full failure | Medium | | Request validation at boundary | Parsing and validating inputs before they reach business logic | High | | Error envelope standardization | Consistent error response format across all endpoints | High | ### Category 3: Concurrency Patterns | Pattern | Problem Solved | Recurrence | |---------|---------------|------------| | Semaphore-based rate limiting | Throttling concurrent operations without blocking the event loop | Medium | | Worker pool with context cancellation | Managing bounded parallelism with graceful shutdown | High | | Read-write lock pattern | Allowing multiple readers but exclusive writers to shared data | Low | ### Category 4: Configuration Patterns | Pattern | Problem Solved | Recurrence | |---------|---------------|------------| | Environment-specific config merging | Layering defaults → env vars → secrets → runtime overrides | High | | Schema-validated config loading | Catching configuration errors at startup rather than at runtime | Medium | | Feature flag rollout strategy | Gradual feature exposure with automated rollback on failure | Medium | --- ## Anti-Pattern Registry Documenting anti-patterns (bad solutions you have witnessed or committed) is equally valuable. Use this template: ```markdown ## [Anti-Pattern Name] — The "X" Pattern **Category:** [problem-domain] **Cost:** [high/medium/low — how much time/money this wastes] ### What It Looks Like ```python # The anti-pattern — recognizable code smell def problematic_code(): """Describe why this is bad.""" pass ``` ### Why It Is Wrong - [Specific reason 1 — e.g., "Creates N+1 database queries"] - [Specific reason 2 — e.g., "No way to handle partial failure gracefully"] - [Specific reason 3 — e.g., ["Makes testing impossible without real dependencies"]] ### Better Alternative Reference your positive pattern entry: see `[Pattern Name]` for the correct approach. ``` --- ## Maintenance Schedule A pattern library that is not maintained becomes a graveyard of stale entries. Follow this schedule: **After Every Task (2 minutes):** - Does the task produce a reusable solution? If yes, add it to your library. - Did you encounter an anti-pattern worth documenting? If yes, add it. **Weekly Review (15 minutes):** - Scan recent additions for duplicates — merge with existing entries if needed - Verify that code examples still work (run them or check imports) - Check that tags are consistent and searchable **Quarterly Cleanup (30 minutes):** - Remove patterns superseded by newer approaches (e.g., "session-based auth" in a JWT-only world) - Archive patterns you have not needed in the past 12 months (move to archive section) - Re-evaluate confidence levels — did your understanding of this pattern deepen? --- ## Pattern Discovery Heuristics Train yourself to spot patterns worth capturing with these mental prompts: **The Second-Time Test:** "Have I dealt with something like this before?" If yes, check your library first. If the answer is "no," consider whether it will be a "before" next time. **The Code Review Reflex:** When reviewing code (yours or others), ask: "Is there a pattern here that should be documented?" Good patterns deserve capture; bad ones deserve anti-pattern documentation. **The Bug Post-Mortem Rule:** Every bug that took more than 30 minutes to debug produces at least one learnable pattern. Extract it. Examples: - Debugging a race condition → "Concurrent state access patterns" - Fixing a CORS issue → "Cross-origin configuration checklist" - Resolving a type mismatch → "Strict typing boundary enforcement" --- ## Constraints ### MUST DO - Document patterns only after validating they have been tested in 2+ scenarios - Include both GOOD and BAD code examples for every pattern entry — the contrast teaches more than either alone - Tag patterns with searchable problem-type keywords, not implementation details - Organize by problem domain, not by language or framework - Reference the `code-philosophy` skill when writing patterns: follow guard clauses, fail fast, use atomic predictability ### MUST NOT DO - Create pattern entries for single-use solutions — this creates noise that drowns out genuine patterns - Write vague descriptions without concrete code examples — "use retries" is not a pattern, "exponential backoff with jitter using these specific parameters" is - Hoard patterns as personal knowledge without sharing when working on a team — the library's value multiplies when shared - Let the library grow to 100+ entries without pruning — stale entries actively harm retrieval accuracy - Organize patterns by language/framework first — this creates silos where "authentication in Python" and "authentication in Go" become separate entries instead of one unified pattern with implementation variants --- ## Output Template When this skill is active and you want to document a new pattern or search existing ones, produce: **For New Pattern Creation:** 1. **Pattern Name** — Clear, searchable title (not language-specific) 2. **Category & Tags** — Problem domain + searchable keywords 3. **Problem Description** — One paragraph with concrete consequences of NOT using this pattern 4. **Code Example** — Full working implementation with type annotations and docstring 5. **Anti-Pattern Counterexample** — The common wrong approach that led you to this pattern **For Pattern Retrieval:** 1. **Best Matching Pattern** — Name, confidence level, and why it matches the current problem 2. **Relevant Variants** — Similar patterns that might be more appropriate depending on context 3. **Known Limitations** — What this pattern does NOT solve (important for setting expectations) --- ## Live References > Authoritative documentation links for this skill's domain. The model follows markdown links at load time to resolve external references and inline content. - [Refactoring.Guru Design Patterns]() - [Clean Architecture (Robert C. Martin)]() - [SOLID Principles of OOP (Wikipedia)]() - [Anti-Patterns in Software Engineering (Wikipedia)]() - [Software Design Patterns Catalog (GoF Book)]() ## Related Skills | Skill | Purpose | |-------|---------| | `personal-workflow-framework` | Your workflow framework tells you WHEN to work; your pattern library tells you HOW, drawing from accumulated experience | | `workflow-patterns` | Generic orchestration patterns — your pattern library is the personalized version capturing your actual solutions | | `skill-creator` | If your patterns are mature enough, they can be formalized into reusable OpenCode skills for broader distribution | > 📖 skill(local cache): code-philosophy