How sathvikar01/agent-engineering-skills works
Agent Engineering Skills is a framework-agnostic collection of 19 reusable SKILL.md guides for designing, building, evaluating, and operating AI agents. The user-to-result workflow is to identify the engineering problem, select the relevant skill, and use its guidance; the README emphasizes loading only the skill that fits. Skills span architecture and workflow, contracts, authority and safety, data, and operations and quality. The README documents these capabilities, but source excerpts are limited to the repository validator, so the graph shows skill guidance rather than internal execution or companion-skill call relationships.
Main entry points: agent-architecture/SKILL.md Agent Architecture · agent-development-workflow/SKILL.md Development Workflow · agent-orchestration/SKILL.md Agent Orchestration · tool-design/SKILL.md Tool Design
Components 20
Design
- Agent Architecture
agent-architecture/SKILL.md - Development Workflow
agent-development-workflow/SKILL.md - Agent Orchestration
agent-orchestration/SKILL.md
Contracts
- Tool Design
tool-design/SKILL.md - Structured Output
structured-output-design/SKILL.md - Prompt Engineering
prompt-engineering/SKILL.md
Authority & Safety
- Deterministic Authority
deterministic-authority/SKILL.md - Human Review
human-in-the-loop/SKILL.md - Agent Guardrails
agent-guardrails/SKILL.md - Agent Security
agent-security/SKILL.md
Data
- RAG Engineering
rag-engineering/SKILL.md - Evidence Provenance
evidence-provenance/SKILL.md - State & Memory
agent-state-and-memory/SKILL.md
Operations & Quality
- Failure Recovery
agent-failure-recovery/SKILL.md - Agent Observability
agent-observability/SKILL.md - Cost & Latency
agent-cost-and-latency/SKILL.md - Agent Evaluations
agent-evals/SKILL.md - Agent Testing
agent-testing/SKILL.md - Tool Evaluations
tool-evals/SKILL.md
Other components
- Agent Builder
Connections 19
- Agent Builder → Agent Architecture: consults for design
README.md - Agent Builder → Development Workflow: consults for lifecycle
README.md - Agent Builder → Agent Orchestration: consults for routing
README.md - Agent Builder → Tool Design: consults for tools
README.md - Agent Builder → Structured Output: consults for output
README.md - Agent Builder → Prompt Engineering: consults for prompts
README.md - Agent Builder → Deterministic Authority: consults for authority
README.md - Agent Builder → Human Review: consults for approval
README.md - Agent Builder → Agent Guardrails: consults for controls
README.md - Agent Builder → Agent Security: consults for threats
README.md - Agent Builder → RAG Engineering: consults for retrieval
README.md - Agent Builder → Evidence Provenance: consults for lineage
README.md - Agent Builder → State & Memory: consults for memory
README.md - Agent Builder → Failure Recovery: consults for recovery
README.md - Agent Builder → Agent Observability: consults for operations
README.md - Agent Builder → Cost & Latency: consults for optimization
README.md - Agent Builder → Agent Evaluations: consults for quality
README.md - Agent Builder → Agent Testing: consults for testing
README.md - Agent Builder → Tool Evaluations: consults for tool quality
README.md