Skip to content

README.md

Latest commit

 

History

History
99 lines (75 loc) · 7.22 KB

README.md

File metadata and controls

99 lines (75 loc) · 7.22 KB
Raw

ADK-Rust Examples

Examples have mostly moved to the dedicated playground repository:

adk-playground — 120+ examples covering agents, tools, workflows, MCP, evaluation, RAG, voice, browser automation, and more.

Also available online at https://playground.adk-rust.com

Local Validation Crates

A small number of live integration crates still live in this repository while their playground versions are being finalized:

  • examples/openai_server_tools — full OpenAI native-tool example matrix covering every exported wrapper
  • examples/anthropic_server_tools — full Anthropic native-tool example matrix for the pinned claudius surface
  • examples/gemini3_builtin_tools — full Gemini native-tool example matrix plus multi-turn mixed-tool validation
  • examples/openai_responses — end-to-end OpenAI Responses validation
  • examples/openrouter — end-to-end OpenRouter validation through the ADK agent stack
  • examples/bedrock_test — Bedrock smoke testing
  • examples/payments — agentic commerce scenario index for ACP/AP2 validation paths
  • examples/developer_ergonomics — developer ergonomics validation (RunnerConfigBuilder, ToolExecutionStrategy, SimpleToolContext, StatefulTool, run_str, #[tool] attributes)
  • examples/acp_client_host — vendor-neutral ACP client with streamed updates, async permissions, and a workspace-bounded filesystem
  • examples/acp_kiro — external coding-agent delegation, persistent sessions, and concurrent cancellation
  • examples/acp_server — expose a tool-using ADK-Rust agent to an editor through stable ACP v1
  • examples/acp_full_protocol — no-API-key, Runner-backed ACP v1 server-direction reference with an end-to-end validating test (embedded-resource + multimodal prompts, permission bridge, session/load replay, usage/tool-call updates)

Validated Feature Examples

Standalone crates demonstrating current ADK-Rust features. Each has its own Cargo.toml, README.md, and .env.example.

No API keys required:

Example Feature Run
examples/agent_registry Agent Registry REST API cargo run --manifest-path examples/agent_registry/Cargo.toml
examples/video_avatar Video Avatar configuration cargo run --manifest-path examples/video_avatar/Cargo.toml
examples/server_builder ServerBuilder + graceful shutdown cargo run --manifest-path examples/server_builder/Cargo.toml

ACP examples:

Example Feature Run
examples/acp_client_host External ACP agent with streamed UI events and client-controlled read-only files Set ACP_AGENT_COMMAND, then cargo run --manifest-path examples/acp_client_host/Cargo.toml
examples/acp_kiro Direct, delegated, persistent, and cancellable coding-agent sessions cargo run --manifest-path examples/acp_kiro/Cargo.toml --bin acp-kiro-session
examples/acp_server Gemini-backed ADK-Rust coding agent exposed to editors Set GOOGLE_API_KEY, then cargo run --manifest-path examples/acp_server/Cargo.toml
examples/acp_full_protocol No-key, Runner-backed ACP v1 server-direction reference + validating test cargo test --manifest-path examples/acp_full_protocol/Cargo.toml

Dry-run mode (no external credentials):

Example Feature Run
examples/mcp_manager Dynamic local MCP server registry with a deterministic Rust fixture cargo run --manifest-path examples/mcp_manager/Cargo.toml
examples/mcp_protocol_revisions MCP 2026-07-28 and SEP-2663 tasks, ending with an LlmAgent; ships its own server cargo run --manifest-path examples/mcp_protocol_revisions/Cargo.toml --bin revisions-agent
examples/slack_toolset Slack Toolset cargo run --manifest-path examples/slack_toolset/Cargo.toml
examples/bigquery_toolset BigQuery Toolset cargo run --manifest-path examples/bigquery_toolset/Cargo.toml
examples/spanner_toolset Spanner Toolset cargo run --manifest-path examples/spanner_toolset/Cargo.toml

Requires GOOGLE_API_KEY:

Example Feature Run
examples/yaml_agent YAML Agent Definition cargo run --manifest-path examples/yaml_agent/Cargo.toml
examples/mcp_sampling Deprecated MCP sampling compatibility cargo build --manifest-path examples/mcp_sampling/Cargo.toml && cargo run --manifest-path examples/mcp_sampling/Cargo.toml --bin sampling-client
examples/secret_provider Secret Provider cargo run --manifest-path examples/secret_provider/Cargo.toml
examples/user_personas User Personas Evaluation cargo run --manifest-path examples/user_personas/Cargo.toml
examples/prompt_optimizer Prompt Optimizer cargo run --manifest-path examples/prompt_optimizer/Cargo.toml
examples/intra_compaction Intra-Compaction cargo run --manifest-path examples/intra_compaction/Cargo.toml
examples/knowledge_graph_agent Knowledge-graph memory for a text agent (remember/relate/load_memory) cargo run --manifest-path examples/knowledge_graph_agent/Cargo.toml
examples/live_translation Real-time speech translation web UI (OpenAI gpt-realtime-translate / Gemini 3.5 Live Translate) cargo run --manifest-path examples/live_translation/Cargo.toml
examples/customer_service Multimodal customer-service voice agent — sees the camera, reads tone, runs refund/handoff tools (OpenAI or Gemini) cargo run --manifest-path examples/customer_service/Cargo.toml

Requires OPENAI_API_KEY:

Example Feature Run
examples/advanced_agents One embedded Runtime gallery for OpenAI chat, AmbientAgent scheduling, Realtime voice, A2A, MCP 2026-07-28 tasks, and telemetry cargo build --manifest-path examples/advanced_agents/Cargo.toml --bins && cargo run --manifest-path examples/advanced_agents/Cargo.toml --bin advanced-runtime
examples/runtime_ui_showcase Embedded runtime UI gallery for a tool-calling agent, graph workflow, and portable team, with screenshots and walkthroughs cargo run --manifest-path examples/runtime_ui_showcase/Cargo.toml --bin runtime-ui-tools
examples/team_architectures Portable TeamSpec examples for handoff, delegation, parallel shared-state swarms, and hybrid workflows cargo run --manifest-path examples/team_architectures/Cargo.toml --bin team-supervisor-handoff

Requires Google Cloud ADC (gcloud auth application-default login):

Example Feature Run
examples/example_store Vertex AI Example Store — few-shot upsert + similarity search against a pre-provisioned store cargo run --manifest-path examples/example_store/Cargo.toml
examples/vertex_sandbox Vertex AI Agent Engine sandbox — create, execute code with input files, delete cargo run --manifest-path examples/vertex_sandbox/Cargo.toml
examples/agent_orchestrator Agent Registry discovery + remote ReasoningEngine delegation cargo run --manifest-path examples/agent_orchestrator/Cargo.toml

Quick Start

git clone https://github.com/zavora-ai/adk-playground.git
cd adk-playground

# Set your API key
export GOOGLE_API_KEY="your-key"

# Run any example
cargo run --example quickstart