# denissergeevitch/agents-best-practices architecture

> Architecture diagram of the GitHub repository denissergeevitch/agents-best-practices, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-04). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/denissergeevitch/agents-best-practices
- Repository: https://github.com/denissergeevitch/agents-best-practices

## Overview

This repository is a provider-neutral agent skill: a compatible agent uses its instructions and reference material to design, generate, audit, refactor, or explain agent harnesses. The documented workflow starts with a user request, activates SKILL.md, and consults topic-specific guidance to produce recommendations or a blueprint. The README names several primary references and lists additional activation topics. Source excerpts were unavailable, so relationships below rely on README statements only; the graph describes guidance, not an implemented runtime harness.

Main entry points: [`SKILL.md`](https://github.com/denissergeevitch/agents-best-practices/blob/main/SKILL.md) (Agent skill), [`references/agentic-loop.md`](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/agentic-loop.md) (Agentic loop), [`references/mvp-agent-blueprint.md`](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/mvp-agent-blueprint.md) (MVP blueprint), [`references/planning-and-goals.md`](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/planning-and-goals.md) (Planning goals)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_core["Skill entry"]
  node_skill["Agent skill<br/>[SKILL.md]"]
end

subgraph group_runtime["Harness design"]
  node_architecture["Architecture<br/>[architecture.md]"]
  node_loop["Agentic loop<br/>[agentic-loop.md]"]
  node_blueprint["MVP blueprint"]
  node_planning["Planning goals"]
  node_memory["Context memory"]
  node_recursive["Recursive harnesses"]
end

subgraph group_tools["Tools and workflows"]
  node_toolsperm["Tool permissions"]
  node_connectors["Skills connectors"]
  node_adaptive["Adaptive tools"]
  node_orchestration["Workflow orchestration"]
  node_teams["Agent teams"]
  node_speculative["Speculative execution"]
  node_provider["Provider patterns"]
end

subgraph group_quality["Evaluation and operations"]
  node_security["Security observability"]
  node_evals["Evaluations<br/>[evals.md]"]
  node_legibility["Feedback loops"]
  node_coverage["Coverage audit<br/>[coverage-audit.md]"]
  node_hardware["Hardware agents<br/>[hardware-agents.md]"]
  node_cost["Prompt caching"]
  node_instructions["System instructions"]
  node_coding["Coding agents<br/>[coding-agents.md]"]
  node_checklists["Checklists<br/>[checklists.md]"]
  node_sources["Source links<br/>[source-links.md]"]
end

node_user(("User"))

node_user -->|"requests guidance"| node_skill
node_skill -->|"guides blueprint"| node_blueprint
node_skill -->|"guides auditing"| node_loop
node_skill -->|"guides auditing"| node_memory
node_skill -->|"guides auditing"| node_security
node_skill -->|"guides auditing"| node_evals
node_skill -->|"guides design"| node_toolsperm
node_skill -->|"guides design"| node_connectors
node_skill -->|"guides adaptation"| node_adaptive
node_skill -->|"covers planning"| node_planning
node_skill -->|"covers workflows"| node_orchestration
node_skill -->|"covers teams"| node_teams
node_skill -->|"covers execution"| node_speculative
node_skill -->|"covers caching"| node_cost

click node_skill "https://github.com/denissergeevitch/agents-best-practices/blob/main/SKILL.md"
click node_architecture "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/architecture.md"
click node_loop "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/agentic-loop.md"
click node_blueprint "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/mvp-agent-blueprint.md"
click node_planning "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/planning-and-goals.md"
click node_memory "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/context-memory-compaction.md"
click node_recursive "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/self-refining-recursive-harnesses.md"
click node_toolsperm "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/tools-and-permissions.md"
click node_connectors "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/skills-and-connectors.md"
click node_adaptive "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/environment-adaptive-tools.md"
click node_orchestration "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/workflow-orchestration.md"
click node_teams "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/adaptive-agent-teams.md"
click node_speculative "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/speculative-tool-execution.md"
click node_provider "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/provider-api-patterns.md"
click node_security "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/security-observability.md"
click node_evals "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/evals.md"
click node_legibility "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/agent-legibility-feedback-loops.md"
click node_coverage "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/coverage-audit.md"
click node_hardware "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/hardware-agents.md"
click node_cost "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/prompt-caching-and-cost.md"
click node_instructions "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/system-prompts-instructions.md"
click node_coding "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/coding-agents.md"
click node_checklists "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/checklists.md"
click node_sources "https://github.com/denissergeevitch/agents-best-practices/blob/main/references/source-links.md"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_skill,node_user toneBlue
class node_architecture,node_loop,node_blueprint,node_planning,node_memory,node_recursive toneAmber
class node_toolsperm,node_connectors,node_adaptive,node_orchestration,node_teams,node_speculative,node_provider toneMint
class node_security,node_evals,node_legibility,node_coverage,node_hardware,node_cost,node_instructions,node_coding,node_checklists,node_sources toneRose
```

## Components

### Skill entry

- [Agent skill](https://github.com/denissergeevitch/agents-best-practices/blob/main/SKILL.md): `SKILL.md`

### Harness design

- [Architecture](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/architecture.md): `references/architecture.md`
- [Agentic loop](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/agentic-loop.md): `references/agentic-loop.md`
- [MVP blueprint](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/mvp-agent-blueprint.md): `references/mvp-agent-blueprint.md`
- [Planning goals](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/planning-and-goals.md): `references/planning-and-goals.md`
- [Context memory](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/context-memory-compaction.md): `references/context-memory-compaction.md`
- [Recursive harnesses](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/self-refining-recursive-harnesses.md): `references/self-refining-recursive-harnesses.md`

### Tools and workflows

- [Tool permissions](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/tools-and-permissions.md): `references/tools-and-permissions.md`
- [Skills connectors](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/skills-and-connectors.md): `references/skills-and-connectors.md`
- [Adaptive tools](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/environment-adaptive-tools.md): `references/environment-adaptive-tools.md`
- [Workflow orchestration](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/workflow-orchestration.md): `references/workflow-orchestration.md`
- [Agent teams](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/adaptive-agent-teams.md): `references/adaptive-agent-teams.md`
- [Speculative execution](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/speculative-tool-execution.md): `references/speculative-tool-execution.md`
- [Provider patterns](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/provider-api-patterns.md): `references/provider-api-patterns.md`

### Evaluation and operations

- [Security observability](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/security-observability.md): `references/security-observability.md`
- [Evaluations](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/evals.md): `references/evals.md`
- [Feedback loops](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/agent-legibility-feedback-loops.md): `references/agent-legibility-feedback-loops.md`
- [Coverage audit](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/coverage-audit.md): `references/coverage-audit.md`
- [Hardware agents](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/hardware-agents.md): `references/hardware-agents.md`
- [Prompt caching](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/prompt-caching-and-cost.md): `references/prompt-caching-and-cost.md`
- [System instructions](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/system-prompts-instructions.md): `references/system-prompts-instructions.md`
- [Coding agents](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/coding-agents.md): `references/coding-agents.md`
- [Checklists](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/checklists.md): `references/checklists.md`
- [Source links](https://github.com/denissergeevitch/agents-best-practices/blob/main/references/source-links.md): `references/source-links.md`

### Other components

- **User**

## Connections

- User → Agent skill: requests guidance (evidence: `README.md`)
- Agent skill → MVP blueprint: guides blueprint (evidence: `README.md`)
- Agent skill → Agentic loop: guides auditing (evidence: `README.md`)
- Agent skill → Context memory: guides auditing (evidence: `README.md`)
- Agent skill → Security observability: guides auditing (evidence: `README.md`)
- Agent skill → Evaluations: guides auditing (evidence: `README.md`)
- Agent skill → Tool permissions: guides design (evidence: `README.md`)
- Agent skill → Skills connectors: guides design (evidence: `README.md`)
- Agent skill → Adaptive tools: guides adaptation (evidence: `README.md`)
- Agent skill → Planning goals: covers planning (evidence: `README.md`)
- Agent skill → Workflow orchestration: covers workflows (evidence: `README.md`)
- Agent skill → Agent teams: covers teams (evidence: `README.md`)
- Agent skill → Speculative execution: covers execution (evidence: `README.md`)
- Agent skill → Prompt caching: covers caching (evidence: `README.md`)
