adk-architecture

Explains how the ADK runtime fits together: the node and graph execution model, Context and Event flow, checkpoint and resume, tracing, and the rules governing the public API surface. Use when answering "how does X work" about ADK internals, tracing where an event or a piece of state comes from, deciding where a new capability belongs, reviewing a change to BaseNode, Workflow, Runner, Agent, Event or Context, working out why a node re-ran or stayed waiting after a resume, or judging whether a change breaks the public API. Don't use for assembling an agent from existing pieces (use adk-agent-builder), diagnosing one failing run or test (use adk-debug), or formatting and naming conventions (use adk-style).

Aller à l'installation Vérifier les fichiers

Dépôt: google/adk-python
Dernière activité de la source: 1 octobre 2026 à 21:28
Langue détectée de SKILL.md: anglais
Étoiles: 21 709
Forks: 4 101

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Installer avec un assistant IA

Copiez ce prompt dans l’assistant IA que vous utilisez.

Follow https://skillsmp.com/skill-install/prompt.md to install only the Agent Skill at https://github.com/google/adk-python/tree/main/.agents/skills/adk-architecture.

Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.

npx skills add https://github.com/google/adk-python --skill adk-architecture

La commande reste sur une seule ligne. Faites défiler horizontalement pour la vérifier avant de la copier.

Vous préférez une copie locale ? Téléchargez les fichiers actuellement disponibles dans SkillsMP.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Explorateur de fichiers

15 fichiers

SKILL.md

Instructions source · Aperçu en lecture seule

name: adk-architecture
description: Explains how the ADK runtime fits together: the node and graph execution model, Context and Event flow, checkpoint and resume, tracing, and the rules governing the public API surface. Use when answering "how does X work" about ADK internals, tracing where an event or a piece of state comes from, deciding where a new capability belongs, reviewing a change to BaseNode, Workflow, Runner, Agent, Event or Context, working out why a node re-ran or stayed waiting after a resume, or judging whether a change breaks the public API. Don't use for assembling an agent from existing pieces (use adk-agent-builder), diagnosing one failing run or test (use adk-debug), or formatting and naming conventions (use adk-style).

# ADK Architecture The runtime is a graph of nodes. `BaseNode` is the unit of execution. `Workflow` is a node that schedules other nodes along declared edges. `NodeRunner` executes exactly one node. `Runner` owns the invocation and the session. Agents are nodes too — `BaseAgent` extends `BaseNode`. A node communicates with its parent through a per-execution `Context`, and with the session through `Event`s it yields. Those are two separate channels: `ctx` carries the result upward, events carry persistence and streaming. Read the source before relying on any signature here. These notes drift; the code does not. Paths below are relative to `src/google/adk/`. ## Pick a reference | Question | Reference | |---|---| | What must a node implement? What may it yield? Which config fields exist? | [BaseNode](references/interface-base-node.md) | | How does the graph schedule nodes, dedup dynamic children, propagate interrupts? | [Workflow](references/interface-workflow.md) | | How does a caller start an invocation? | [Runner](references/interface-runner.md) | | What is `Agent`, and which methods do I call on it? | [Agent](references/interface-agent.md) | | I am subclassing an agent — what do I override? | [BaseAgent](references/interface-base-agent.md) | | What is on an `Event`, and what may I assume about its lifetime? | [Event](references/interface-event.md) | | What does a node read and write on `ctx`? | [Context](references/architecture-context.md) | | Who creates the child Context, stamps events, retries, catches errors? | [NodeRunner](references/architecture-node-runner.md) | | Why are Runner, NodeRunner and Workflow three separate things? | [Runner roles](references/architecture-runner-roles.md) | | How does a human-in-the-loop pause and resume work for one node? | [Checkpoint and resume](references/architecture-checkpoint-resume.md) | | How does a whole workflow survive a pause, and what does `is_resumable` change? | [Workflow resumability](references/architecture-workflow-resumability.md) | | How are spans created, and what attributes do they carry? | [Observability](references/architecture-observability.md) | | Why does the model not see the raw event log? | [LLM context orchestration](references/architecture-llm-context-orchestration.md) | | Is this change a breaking change? Where does a new export belong? | [API principles](references/api-principles.md) | ## Where the code lives | Concept | Module | |---|---| | `BaseNode`, `START` | `workflow/_base_node.py` | | `Workflow`, `_LoopState` | `workflow/_workflow.py` | | `Graph`, edge compilation | `workflow/_graph.py` | | `NodeRunner` | `workflow/_node_runner.py` | | `DynamicNodeScheduler`, `DynamicNodeState`, `DynamicNodeRun` | `workflow/_dynamic_node_scheduler.py` | | `ReplayManager` (resume scan) | `workflow/utils/_replay_manager.py` | | `check_interception` (replay fast-forward) | `workflow/utils/_replay_interceptor.py` | | `NodeInterruptedError`, `NodeTimeoutError` | `workflow/_errors.py` | | `Context`, `ctx.run_node()` | `agents/context.py` | | `ReadonlyContext` | `agents/readonly_context.py` | | `InvocationContext` | `agents/invocation_context.py` | | `BaseAgent`, `LlmAgent` (aliased `Agent`) | `agents/base_agent.py`, `agents/llm_agent.py` | | `BaseLlmFlow`, `SingleFlow`, `AutoFlow`, `flows/llm_flows/` subpackages (`context/`, `prompt/`, `tools/`, `extensions/`, `core/`) | `flows/llm_flows/` | | `Event`, `NodeInfo` | `events/event.py` | | `EventActions` | `events/event_actions.py` | | Branch paths (`parent.child@1`) | `events/_branch_path.py` | | Node paths (`wf@1/child@2`) | `events/_node_path_builder.py` | | `Runner`, `InMemoryRunner` | `runners.py` | | `LiveRequestQueue`, `LiveRequest` | `live/live_request_queue.py` | | Node spans, `TelemetryContext` | `telemetry/node_tracing.py` | | `ResumabilityConfig` | `apps/_configs.py` | Everything under `workflow/` is a leading-underscore module. Treat those names as internal — they can change without a major version bump, so a change there is not automatically a breaking change.

Voir sur GitHub

Retour à la recherche Retour à l’installation

adk-review

Reviews the uncommitted changes in an adk-python working tree and reports correctness, design, public-API stability, test, sample and documentation gaps as a prioritized findings report, fixing them only when asked. Use when the user asks to review local changes, wants a self-review before opening a pull request, asks whether a change breaks the public API or needs tests, samples or docs, or asks what is wrong with the current diff. Required for changes to public APIs, core architecture (Runner, Workflow, BaseNode), new features and major refactors. Don't use for a single style nit (use adk-style), for diagnosing a failing test or a misbehaving agent at runtime (use adk-debug), or for wording a commit message or PR description (use adk-git).

2026-10-03

adk-style

Python style and codebase conventions for ADK (Agent Development Kit): private-by-default file visibility, imports, type hints, Pydantic v2 models, formatting, docstrings, logging, async I/O, file and test layout, and unit test structure. Use when writing or editing ADK source or tests, deciding whether a new file or symbol should be public or private, naming or placing a test file, fixing a formatter, linter, or type-check failure (pyink, isort, ruff, mypy, addlicense, compliance-checks), or asking whether code matches house style. Don't use for reviewing a whole changeset (use adk-review), writing a developer guide or design doc for a code unit (use adk-unit-guide or adk-unit-design), building or configuring agents (use adk-agent-builder), or installing the toolchain (use adk-setup).

2026-10-03

adk-agent-builder

Builds ADK (Agent Development Kit) Python agents: LLM agents with tools, graph workflows of function and agent nodes, conditional routing, fan-out and join, schema-validated delegation between agents, human-in-the-loop pauses, and pytest coverage for all of it. Use when asked to create an agent or a workflow, add a tool to one, branch or loop between nodes, run steps in parallel, pause for user approval, or test an agent. Don't use for explaining how ADK works internally or designing its core components (use `adk-architecture`), for an agent that already runs but misbehaves (use `adk-debug`), for authoring a sample under `contributing/` (use `adk-sample-creator`), or for naming, typing, and formatting conventions (use `adk-style`).

2026-10-01

Tout voir dans ce dépôt

adk-architecture

Informations de source

Options d'installation

Installer avec un assistant IA

Vérifiez les fichiers source

Plus depuis ce dépôt

Installer avec un assistant IA

Plus depuis ce dépôt