Autonomous Agent Harness
Building autonomous AI agent workflows with bounded autonomy, guardrails, tool calling, self-correction, escalation patterns, safety constraints, output validation, and sandboxing.
Git Workflow
Git workflow for branches, commits, merges, and conflict management.
Exa Search
Exa search — MCP server usage, semantic search queries, query optimization, result filtering, relevance scoring, research workflows, comparing results across queries, combining Exa with Context7 docs lookup, search result validation.
Frontend Design
Workflow guidance for Frontend Design.
Postgres Patterns
Workflow guidance for Postgres Patterns.
Claude Api
Claude API usage — messages API, system prompts, tool use/function calling, extended thinking, prompt caching, token counting, streaming, image understanding, structured output, rate limits, fallback strategies, cost optimization.
Security Review
Security review methodology covering OWASP Top 10, input validation audit, auth/authorization review, secret management check, dependency CVE scan, threat modeling, and secure code review checklist.
Context Prune
Context window management: context budgeting, priority-based pruning, conversation summarization, reference replacement, token counting, strategic truncation, and context window optimization for Claude/GPT models.
Ci Pipeline
GitHub Actions, Jenkins pipelines, PR automation, and quality gates.
Backend Patterns
FastAPI architecture patterns including system design, dependency injection, error handling, background tasks, WebSocket, testing, middleware, database integration, and OpenAPI design.
Python Testing
Python testing workflow covering Pytest, FastAPI testing, Factory Boy, mocking, integration testing, property-based testing with Hypothesis, coverage, xdist parallelization, and AI-focused regression testing.
Search First
Search-first methodology — search before coding discipline, effective search strategies, when to search vs reason, query formulation, result verification, avoiding stale information, cross-referencing, documenting search findings.
Langchain Llamaindex
LangChain and LlamaIndex patterns for RAG pipelines, agent workflows, retrieval strategies, and cost-optimized LLM chains. Covers advanced RAG, multi-hop retrieval, reranking, and structured output.
Api Integrations
API connector patterns for OAuth lifecycle, webhooks, rate limiting, and agent-native tool calling.
Data Pipelines Ai
Data pipelines for AI workflows — Airflow DAGs, Prefect flows, Kafka streaming, batch vs streaming, data versioning, chunking, idempotent design, dataset curation, and pipeline monitoring.
Coding Standards
Project-wide coding standards enforcement: naming conventions, file organization, function size limits, complexity budgets, style guides, linting rules, code review standards, and language-specific conventions for Python and TypeScript. Includes linter configs and rules.
Api Design
RESTful API design — resource naming, error envelopes, versioning, pagination strategies, rate limiting, idempotency keys, request/response validation with Pydantic/Zod, HATEOAS, OpenAPI generation, API documentation.
Pytorch Patterns
PyTorch patterns — nn.Module model definition, training loops, DataLoader optimization, GPU utilization, distributed training with DDP, torch.compile, mixed precision, gradient accumulation, checkpointing, learning rate scheduling, CUDA toolchain.
Python Patterns
Python data and backend patterns including Pandas, NumPy, SQLAlchemy 2.0, Alembic, data validation, performance optimization, and OOP composition.
Codebase Report
Structured codebase reporting skill. Combines outputs from codebase-analyzer and codebase-learner agents into a unified report covering architecture, data flow, coding patterns, gotchas, confidence ratings, and recommended study paths. Used when onboarding new developers or auditing an unfamiliar codebase.
Backlog Management
Manage the feature backlog from docs/features/ and Git issues. Prioritizes features, tracks readiness, identifies specs missing details, and surfaces what needs human input. Used by the multi-agent-project-manager agent.
Frontend Patterns
React 19 and Next.js patterns including component architecture, state management, performance, styling, forms, testing, Chrome extension parity, DX tooling, micro-frontends, UI state machines, and AI chat surface composition.
Eval Harness
Evaluation harness for LLM and code quality assessment. Covers Pass@k metrics, golden datasets, LLM-as-judge, benchmark suites, regression detection, and eval-driven development workflow.
Continuous Agent Loop
Continuous agent loop patterns: bounded automation loops, quality-gated iteration, loop health monitoring, termination criteria, human-in-the-loop gates, loop state management, failure recovery, and idempotent operations.
Architecture Decision Records
ADR process — template, lifecycle (proposed → accepted → deprecated), storage convention, when to create ADRs, referencing ADRs in PRs and plans, ADR maintenance.
Observability Telemetry
Observability & telemetry — OpenTelemetry SDK/collector, Prometheus metrics, Grafana dashboards, structured logging, distributed tracing, RAG span attributes, event stream monitoring, Sentry integration, chaos drills, and automation.
Openai Api
OpenAI API production patterns — Completions, Embeddings, function calling, fine-tuning, Batch API, streaming, Assistants API, cost optimization, error handling, safety, evaluation, and production architecture.
Mlops Rag
MLOps for RAG systems — model versioning, A/B testing, embedding pipelines, vector DB comparison, evaluation, observability, and continuous delivery for retrieval-augmented generation.
Prompt Optimizer
Prompt optimization — compression, few-shot example selection, chain-of-thought alternatives, structured output, system prompt design, versioning and A/B testing, injection defenses, evaluation metrics, cache optimization, distillation patterns.
Code Review
Code review methodology for PRs covering correctness, security, performance, and maintainability. Includes DIFF analysis, risk assessment, test coverage verification, and a comprehensive review checklist.
Document Processing
Parse, clean, validate, and convert PDFs, spreadsheets, CSVs, and documents into structured data.
Codebase Onboarding
Codebase onboarding methodology: dependency mapping, architecture discovery, key file identification, domain terminology extraction, first task identification, learning checklist, and knowledge gap tracking. Includes scripts for automated codebase analysis.
Multi Agent Git Workflow
Multi-agent Git workflow for parallel AI development — repo scanning, micro-task breakdown, dynamic agent spawning (backend-1, backend-2, frontend-1, etc.), continuous test/eval loop, AI-as-judge validation, automated PRs, quality gates. Stops ONLY when feature is complete: all tests pass, evals green, lint/format/type-check clean.
Data Pipelines
Data pipeline workflow for ETL, validation, and quality controls.
Iterative Retrieval
Iterative retrieval — progressive context loading, retrieval-then-refine loops, query rewriting, adaptive top-k, multi-hop retrieval, confidence scoring, fallback for low-confidence results, retrieval-time prompt guards.
E2e Testing
Playwright E2E testing patterns including page object model, fixtures, parallel execution, flaky test prevention, visual regression, API mocking, and cross-browser testing.
Aws Deployment
AWS deployment patterns: ECS Fargate, Lambda, EC2, App Runner, CloudFormation, CDK. Infrastructure as code, blue-green deployments, rolling updates, health checks, auto-scaling, CloudWatch integration.
Mcp Server Patterns
MCP server patterns — server configuration, tool definition, resource exposure, prompt template integration, server security, multi-server orchestration, context passing, error handling, MCP protocol basics.
Github Ops
Workflow guidance for Github Ops.
Aws Devops
AWS DevOps — ECS, Lambda, Docker, Kubernetes, ArgoCD, GitHub Actions, Jenkins, Ansible, CI/CD pipelines, IaC, monitoring, cost optimization, and security.
Agentic Engineering
Agentic engineering patterns — agent orchestration, delegation contracts, multi-agent system design, memory graph usage, streaming/async pipelines, OOP decomposition, hierarchical delegation, context window management, and agent failure recovery.
Deployment Patterns
Workflow guidance for Deployment Patterns.
Database Migrations
Workflow guidance for Database Migrations.
Autonomous Loops
Autonomous loop patterns: self-improving workflows, continuous feedback loops, quality gates, loop termination conditions, retry strategies, bounded iteration, and convergence detection.
Docker Patterns
Workflow guidance for Docker Patterns.
Documentation Lookup
Workflow guidance for Documentation Lookup.
Dmux Workflows
Workflow guidance for Dmux Workflows.
Ai Dev Kit
Production-ready AI dev workflow scaffold with 59 skills, 33 agents, hooks, and rules for Claude Code, Codex, Gemini CLI, Copilot CLI, Qwen Code, and OpenCode.
Mlops Workflow
MLOps workflow for training, evaluation, deployment, and retraining.
Hexagonal Architecture
Workflow guidance for Hexagonal Architecture.
Deep Research
Deep research methodology — multi-source investigation, web search with Exa, documentation traversal, cross-referencing sources, synthesizing findings, research report structure, evidence grading, conflicting source resolution, citation format.