agents-build

An agent skill by aws, from aws/agent-toolkit-for-aws. Tags: ai, architecture, automation, integration, multi-agent.

What it does

Use to extend an existing agent project with memory, app integration, VPC, multi-agent, migration, model, browser, code interpreter, payments, or resource removal. Triggers: "add memory", "remember across sessions", "call agent from app", "invoke agent from code", "agent auth", "streaming", "VPC", "VPC connectivity", "can't reach from VPC", "multi-agent", "A2A", "A2A auth", "orchestrator not delegating", "specialist not called", "migrate Bedrock Agent", "migration issue", "change model", "browser tool", "code interpreter", "delete agent", "tear down", "agentcore remove", "cross-account memory", "add payments capability to my agent", "wire payments plugin", "integrate x402 payments with the agent I'm building", "add MPP payments", "Machine Payments Protocol". External APIs via Gateway: use agents-connect. New project: use agents-get-started. CLI/dev-server errors: use agents-debug. Runtime x402/MPP payments: use agents-pay. Migration-specific Strands vs LangGraph routes here.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add aws/agent-toolkit-for-aws --skill agents-build

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/aws/agent-toolkit-for-aws
cp -r agent-toolkit-for-aws/plugins/aws-agents/skills/agents-build ~/.claude/skills/agents-build

Safety box score

Grade D, elevated risk: 58 of 100 risk points, rated 2026-10-20. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a broad AgentCore project builder that can add memory, browser/code-interpreter, VPC, multi-agent, and payment capabilities, and it also includes teardown flows. The main risks are destructive AWS cleanup, broad network/account changes, and helper scripts that make outbound HTTP/payment calls and read/write local credential-related files.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	It ships runnable Python helpers and shell-driven workflows (`scripts/process_payment_tool.py`, `agentcore deploy`, `aws ...`), so using the skill can execute code and commands.
Network access	2 (moderate)	It talks to AWS APIs, agent endpoints, browser sessions, and arbitrary HTTPS URLs (`httpx.request`, `curl http://localhost:9000`, `aws ...`), so data and requests leave the machine.
Filesystem reach	1 (low)	It mainly reads and writes project-local files like `agentcore/agentcore.json`, `app/<AgentName>/main.py`, and `agentcore/.env.local`, with no system-path editing.
Secrets & credentials	2 (moderate)	It explicitly handles credentials and tokens via `agentcore/.env.local`, `@requires_api_key`, `@requires_access_token`, and payment-related environment variables.
Privilege & persistence	2 (moderate)	It changes AWS account and network settings (`create-nat-gateway`, VPC config, IAM policies, deploy/destroy flows), which requires elevated cloud privileges even though it does not use local sudo.
Instruction hijack surface	1 (low)	The docs are imperative and agent-directed, but I saw no hidden instructions, remote prompt fetching, or explicit attempts to override higher-priority rules.
Irreversible actions	3 (high)	The skill can deploy/update agents, create/delete cloud resources, open payment sessions, and tear down infrastructure, including irreversible delete flows like `delete-stack` and `delete-repository --force`.
Transparency	1 (low)	One shipped reference, `references/payments.md`, could not be read, so the payment workflow was not fully reviewable.

Before installing

Read `references/teardown.md` first if you might remove resources; it contains destructive delete commands.
Review `scripts/process_payment_tool.py` and `scripts/setup_payment_user.py` if you do not want outbound HTTP or payment-session setup.
Check whether `agentcore/.env.local` and `agentcore/.cli/deployed-state.json` will be created or modified on your machine.
Only install if you are comfortable granting AWS permissions for VPC, IAM, deploy, and destroy operations.

Rated 2026-10-20 by an automated review of 13 files from the skill's repository.

Source

Repository: aws/agent-toolkit-for-aws (all skills from this repository)
Path: plugins/aws-agents/skills/agents-build/SKILL.md
Branch: main
Collection: aws-agents
Updated: 2026-09-19

orchestrate — Use only when the user explicitly types `/orchestrate <goal>` to decompose a large task.
ralph-loop — Start a Ralph Loop for iterative self-referential development.
agents-connect — Use when connecting your agent to external APIs, tools, or services via Gateway, or restricting tool access with Cedar policies.
aws-blocks — Guides building full-stack applications with AWS Blocks — an Infrastructure-from-Code framework.
aws-messaging-and-streaming — Guides general use of AWS messaging and streaming services.
coordinating-multi-space-devops-agent — Coordinate the AWS DevOps Agent across multiple AgentSpaces from one Claude Code session — route questions to the right space (prod vs staging vs knowledge).