PatternsReal-world Examples

Marketing Plan & Implement Agent

Two-phase marketing agent with brand context: Stage 0 extracts brand info from URL or manual input, planning phase gathers requirements and creates a plan with todos, implementation phase executes the plan using web scraping and research tools. Uses callOptionsSchema + prepareCall for dynamic agent configuration.

Loading preview…

Problems solved

Technical challenges this implementation handles out of the box.

Extract brand context from a URL or manual intake
Plan marketing work with human-approved todos
Execute the plan with research and scraping tools
Pass brandContext at runtime via callOptions and prepareCall
Share runtime context with tools through experimental_context

Use cases

Products and workflows this pattern is designed to support.

Marketing campaign plan-and-execute agents
Competitor research implementation loops
Content strategy agents with approval gates
SEO research that moves from plan to action

Setup

Requirements, wiring steps, and what this pattern adds to your project.

Getting started

Pick how you want to pull this pattern in — then wire env vars and routes the same way.

1
Install from the preview toolbar
Copy the install command above and run it in your project — Pro patterns include a short-lived token.
2
Add environment variables to .env.local
OPENAI_API_KEY, FIRECRAWL_API_KEY
3
Point client fetches at your API routes
Replace preview `/view/.../api/...` paths with `/api/...` in the installed files.
4
Customize the agent and tool files
Adapt prompts, tools, and stop conditions for your product — Copy for AI in the toolbar helps seed that work.
5
Run your dev server and open the pattern route
Install dependencies if needed, then start the app and verify responses.

Environment variables

OPENAI_API_KEYGet key
FIRECRAWL_API_KEYGet key

Capabilities

AI SDK APIs

useChatgatewayToolLoopAgenttoolgenerateObjecthasToolCallstepCountIscreateAgentUIStreamResponseInferAgentUIMessagecallOptionsSchemaprepareCallexperimental_context

Providers

OpenAI

External services

Firecrawl

Files added

24 files

app/page.tsx
app/layout.tsx
app/api/agent-usecase-marketing-plan/route.ts
app/api/agent-usecase-marketing-implement/route.ts
lib/marketing-plan-types.ts
lib/marketing-plan-actions.ts
lib/cheerio-scraper.ts
lib/utils/format-brand-context.ts
lib/ai/plan/plan-agent.ts
lib/ai/plan/tools/index.ts
lib/ai/plan/tools/ask-multiple-choice.ts
lib/ai/plan/tools/ask-text-input.ts
lib/ai/plan/tools/generate-plan.ts
lib/ai/plan/tools/refine-plan.ts
lib/ai/implement/implement-agent.ts
lib/ai/implement/tools/index.ts
lib/ai/implement/tools/read-plan.ts
lib/ai/implement/tools/update-plan.ts
lib/ai/implement/tools/read-todo.ts
lib/ai/implement/tools/update-todo.ts
lib/ai/implement/tools/finalize.ts
lib/ai/implement/tools/firecrawl.ts
lib/ai/implement/tools/url-scraper.ts
README.md

Dependencies

18 total

npm packages

AISDKcheeriofirecrawl-aisdklucide-reactNext.jsReactsonnerstreamdownuse-stick-to-bottomZod

Registry components

buttoncollapsibleemptyinputtextareatooltiptabslabel

Critical files

agent-usecase-marketing-plan/route.ts
Plan phase API route using createAgentUIStreamResponse with callOptions to pass brandContext at runtime.
agent-usecase-marketing-implement/route.ts
Implement phase API route passing both brandContext and plan via callOptions.
plan/plan-agent.ts
Plan builder ToolLoopAgent with callOptionsSchema + prepareCall for runtime brand context injection into instructions and experimental_context.
tools/ask-multiple-choice.ts
Brand-context-aware multiple choice tool. Reads brandContext from experimental_context to tailor AI-generated questions.
tools/generate-plan.ts
Plan generation tool with brand context awareness using generateObject with planSchema.
tools/refine-plan.ts
Plan refinement tool with brand-context-aware iteration.
tools/ask-text-input.ts
Brand-context-aware text input tool for gathering open-ended requirements.
implement/implement-agent.ts
Autonomous implementation agent with prepareStep for phased workflow (triage, process todos, research, synthesis, finalize). Most advanced agent pattern: callOptionsSchema, prepareCall, prepareStep, stopWhen.
tools/finalize.ts
Implementation finalization tool that reads plan from experimental_context and signals completion.
tools/update-plan.ts
Tool for updating plan sections from experimental_context.
tools/update-todo.ts
Tool for updating todo status with validation against plan.
tools/read-plan.ts
Tool for reading current plan state from experimental_context.
tools/read-todo.ts
Tool for reading/filtering todos by ID or status from the plan context.
tools/firecrawl.ts
Firecrawl web search tool for real-time web research during plan implementation.
lib/marketing-plan-types.ts
Complete type system including Zod schemas (BrandContext, Plan, Todo) and all type definitions for both plan and implement phases.
utils/format-brand-context.ts
Utility that formats BrandContext into markdown for injection into agent instructions.