adk-agent-builder

Build production-ready AI agents using Google's Agent Development Kit with AI assistant integration, React patterns, multi-agent orchestration, and comprehensive tool libraries. Use when appropriate context detected. Trigger with relevant phrases based on skill purpose.

Install with Tessl CLI

npx tessl i github:jeremylongshore/claude-code-plugins-plus-skills --skill adk-agent-builder

What are skills?

1.31x

Quality

37%

Does it follow best practices?

Impact

63%

1.31x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./plugins/jeremy-google-adk/skills/adk-agent-builder/SKILL.md

SKILL.md

Review

Evals

Evaluation results

64%

22%

News Research Agent: Initial Project Setup

Single ReAct agent scaffold structure

Criteria

Without context

With context

src/agents/ directory

25%

100%

src/tools/ directory

25%

100%

tests/ directory

75%

100%

Dependency file present

100%

google-adk in dependencies

100%

anthropic>=0.18.0 in dependencies

100%

ReactAgent base class

ToolRegistry usage

No hardcoded secrets

100%

Smoke test present

25%

Regression test for tool failure

Tool stubs / wiring points

100%

claude-3-5-sonnet-20241022 model

Without context: $0.7541 · 7m 10s · 40 turns · 37 in / 8,229 out tokens

With context: $1.1899 · 7m 19s · 47 turns · 364 in / 12,553 out tokens

46%

Company Data Enrichment Tool for Sales Agent

Custom tool implementation and rate-limit resilience

Criteria

Without context

With context

Tool base class

ToolResult returned

input_schema property

Exponential backoff

100%

max_retries=3

base_delay=1

100%

RateLimitError handled

100%

Structured logging

Regression test for tool failure

100%

No hardcoded secrets

100%

ToolResult success field

Without context: $0.8244 · 7m 33s · 36 turns · 38 in / 11,868 out tokens

With context: $1.1482 · 7m 20s · 41 turns · 41 in / 11,489 out tokens

80%

24%

Document Processing Pipeline: Multi-Agent Setup and Cloud Deployment

Multi-agent orchestration and Vertex AI deployment

Criteria

Without context

With context

MultiAgentOrchestrator used

30%

50%

Specialized sub-agents

100%

adk deploy command

100%

AgentCard/task endpoint check

37%

100%

Permissions check in checklist

100%

Logs check in checklist

100%

Least-privilege auth guidance

100%

Pinned install commands

25%

Stateless agent design

25%

37%

No hardcoded secrets

100%

Dependency file present

50%

100%

src/ directory structure

83%

100%

claude-3-5-sonnet-20241022 model

Without context: $0.7602 · 7m 21s · 39 turns · 36 in / 10,781 out tokens

With context: $1.9143 · 8m 45s · 56 turns · 54 in / 23,458 out tokens

Evaluated: 5 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

News Research Agent: Initial Project Setup Company Data Enrichment Tool for Sales Agent Document Processing Pipeline: Multi-Agent Setup and Cloud Deployment

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.