Skip to content

Repository files navigation

Backend Engineering Intelligence

Backend engineering intelligence for coding agents. A reusable skill system for auditing, architecting, debugging, securing, optimizing, and production-hardening backend systems using curated engineering knowledge, deterministic inspection, architecture patterns, evaluation rubrics, and stack-specific recipes.

Pi Agent TypeScript Node.js Python PostgreSQL SQLite Redis License: MIT


Supported Ecosystem & Stacks

Framework / Language Persistence & Data Async & Messaging Agent Runtimes
Node.js (Express, Fastify, NestJS), Python (FastAPI, Django, Flask), Go, Rust PostgreSQL, SQLite, DuckDB, MySQL, Redis, S3 / Object Storage Redis BullMQ, Celery, RabbitMQ, Kafka, AWS SQS Pi Agent, Hermes, Codex, Claude Code

Overview

AI coding agents excel at producing functional endpoints, but frequently introduce backend "slop": premature microservices, unindexed database tables, missing transaction boundaries, silent exception handling, wildcard CORS headers, unvalidated user inputs, and missing idempotency rules.

backend-engineering-intelligence equips AI coding agents with the pragmatic judgment and workflows of senior backend architects, principal database engineers, and production reliability engineers. It guides agents to build software that is correct, secure, reliable, performant, and maintainable without turning them into "architecture astronauts".


Design Philosophy

The core priority ranking for all backend intelligence recommendations is:

CORRECTNESS > DATA INTEGRITY > RELIABILITY > SECURITY > OBSERVABILITY > MAINTAINABILITY > PERFORMANCE > SCALABILITY > ARCHITECTURAL NOVELTY

Key Operating Principles

  1. Pragmatic Monolith First: Keep services together until independent deployment or team boundaries demand splitting.
  2. Evidence Before Cache: Measure SQL query latency before adding Redis or caching layers.
  3. Data Boundary Integrity: Validate all input data at system boundaries; enforce constraints at the database layer.
  4. Resilience Over Magic: Idempotency rules, timeouts, and transaction limits precede complex retry or queue orchestration.
  5. No Architecture Astronauts: Reject enterprise boilerplate, empty interface abstractions, and premature microservices.

Core Pi Agent Skills

Skill Command Role Description Modifies Code? Best Used For
/skill:backend-director Orchestrator Master intent-translation and backend director skill. Translates vague prompts ("Bro this backend is messy") into structured engineering briefs and routes work to specialists. Yes Vague or informal backend refactoring, production prep, or performance requests.
/skill:backend-audit Specialist Non-destructive backend system assessment. Scores project across 14 rubric dimensions and flags anti-slop violations. No Code reviews, architectural health checks, and pre-release audits.
/skill:api-architect Specialist API design and contract specialist. Designs REST, RPC, GraphQL contracts, schema validations, pagination, error models, and idempotency key handling. Yes Endpoint design, schema validation, contract stability, and API refactoring.
/skill:data-layer Specialist Database and persistence specialist. Designs schemas, migrations, SQL queries, indexes, connection pools, transaction boundaries, and ORM usage. Yes Schema design, N+1 query elimination, indexing hot paths, and migration safety.
/skill:reliability-pass Specialist Production reliability specialist. Audits and applies timeouts, retries, idempotency, circuit breakers, graceful shutdown, and health checks. Yes Preventing cascade failures, worker queue crashes, and zombie connection leaks.
/skill:security-pass Specialist Practical application security review. Checks AuthN/AuthZ, secrets management, input validation, SQLi, SSRF, CORS, path traversal, and LLM tool safety. Yes Hardening endpoints, secret leakage prevention, and input sanitization.
/skill:performance-pass Specialist Evidence-based performance engineering. Establishes baseline measurement → profiles SQL/serialization → eliminates bottlenecks → benchmarks again. Yes Latency reduction, high throughput tuning, memory leak fixes, and payload optimization.
/skill:production-readiness Specialist Last-20%-before-launch specialist. Verifies environment configs, migrations, health checks, structured logging, secrets, backups, rollback plans, and smoke tests. Yes Preparing prototypes for live production demos, client deployments, and scaling.

14-Dimension Backend Quality Rubric

Projects are evaluated out of 100 points across 14 dimensions defined in evals/rubric.md:

  1. Architectural Coherence (/10)
  2. Correctness (/10)
  3. API Contract Quality (/10)
  4. Data Integrity (/10)
  5. Security (/10)
  6. Reliability (/10)
  7. Error Handling (/10)
  8. Observability (/10)
  9. Performance (/10)
  10. Scalability Appropriateness (/10)
  11. Testing (/10)
  12. Deployment Readiness (/10)
  13. Developer Experience (/10)
  14. Maintainability (/10)

Critical Failure Gates

A high score in general dimensions cannot mask critical risks. The audit automatically issues a BLOCKER verdict if any of the following exist:

  • Broken or bypassable Authentication/Authorization
  • Data Loss / Unsafe Destructive Schema Migration
  • SQL Injection or Remote Code Execution vulnerability
  • Hardcoded Secrets / Credentials in Source Control
  • Missing Transaction Boundaries on Multi-Write Operations
  • Unhandled HTTP 500 exceptions on primary user flows

Installation & Usage

1. Global Installation (Recommended)

Makes the backend intelligence package available across all Pi Agent projects on your system:

pi install git:github.com/kerwinarlan/backend-engineering-intelligence

2. Project-Local Installation

Adds the skills to the current project's .pi/settings.json:

pi install git:github.com/kerwinarlan/backend-engineering-intelligence -l

3. Quick Start Examples

Natural Language Request (Backend Director)

/skill:backend-director

Bro this backend is super messy. Requests are timing out when DB gets loaded, and I'm getting random 500s on checkout. Help me get it production ready.

Non-Destructive Audit

/skill:backend-audit

CLI Audit Tool

Run static anti-slop analysis and security checks against any backend codebase:

npm run audit -- /path/to/target-backend

Repository Structure

backend-engineering-intelligence/
├── README.md
├── AGENTS.md
├── LICENSE
├── package.json
├── tsconfig.json
│
├── skills/                     # 8 Canonical Pi Agent Skills
│   ├── backend-director/       # Master Intent-Translation & Orchestrator
│   ├── backend-audit/          # Non-destructive 14-point audit
│   ├── api-architect/          # REST/RPC/GraphQL contract design & schema validation
│   ├── data-layer/             # Schemas, SQL, migrations, indexing, transactions
│   ├── reliability-pass/       # Timeouts, retries, idempotency, graceful shutdown
│   ├── security-pass/          # AuthN/AuthZ, OWASP Top 10, LLM tool boundaries
│   ├── performance-pass/       # Baseline -> Profile -> Bottleneck -> Benchmark
│   └── production-readiness/   # Env config, health checks, logging, rollback
│
├── .pi/                        # Local Project Config
│   ├── settings.json
│   └── skills -> ../skills     # Symlink to skills/
│
├── docs/                       # Architectural Specifications
│   ├── ARCHITECTURE.md         # Progressive disclosure design
│   ├── ANTI_SLOP.md            # Anti-overengineering rules across 9 backend domains
│   ├── DESIGN_PHILOSOPHY.md    # Core priorities & decision rules
│   ├── TAXONOMY.md             # Categorization of backend patterns
│   └── CONTRIBUTING_KNOWLEDGE.md
│
├── knowledge/                  # Curated Engineering Knowledge
│   ├── architecture/           # Modular monoliths, service boundaries, jobs
│   ├── api/                    # REST, pagination, idempotency, versioning
│   ├── data/                   # Transactions, schema, query plans, connection pools
│   ├── reliability/            # Timeouts, retries, health/readiness probes
│   ├── security/               # Auth, validation, injection, secrets, LLM safety
│   ├── observability/          # Structured logs, metrics, tracing, correlation IDs
│   ├── performance/            # Profiling, indexing, caching strategies, serialization
│   ├── testing/                # Integration tests, migration tests, failure injection
│   ├── deployment/             # Twelve-factor, health probes, graceful shutdown
│   ├── ai-backends/            # LLM boundaries, prompt injection, token limits
│   ├── contexts/               # SaaS, AI Agents, Data Pipelines, Civic-Tech, Mobile
│   ├── recipes/                # Express, FastAPI, Prisma, SQLAlchemy, Redis
│   └── references/patterns/    # JSON reference catalog of proven backend patterns
│
├── evals/                      # Audit Rubric & Test Harness
│   ├── rubric.md               # 14-Dimension Rubric & Critical Gates
│   ├── eval-fixture.ts         # Evaluation harness runner
│   └── fixtures/               # Sample anti-pattern vs clean fixtures
│
└── scripts/                    # Deterministic Inspection Tooling
    ├── validate-skills.ts      # Skill schema validator
    ├── validate-knowledge.ts   # Knowledge base validator
    ├── audit-project.ts        # CLI backend anti-slop audit engine
    └── db-inspector.ts         # Schema & migration inspector

License

MIT License © 2026 Kerwin Arlan

About

Backend engineering intelligence for coding agents. Audit, architect, debug, secure, optimize, and production-harden APIs, databases, workers, and services with reusable Pi Agent skills.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages