🎁 Not sure yet? Get the free sample pack →
← All packs

Agent Builder

System prompts, zod tool schemas, guardrails and eval cases for 15 agents, a design guide, and AI SDK v7 and OpenAI Agents SDK examples that compile. Replaces Agent System Prompt Starters.

Version 1.0.0 · 27 files · 5,527 words · instant zip download

$12Buy for $12By purchasing you agree to the Terms.

14-day refund if the pack didn't help you, one refund per customer. Reply to your receipt email to ask.

Or get all packs for $39 (regular $61): All Packs Bundle.

Included files

  • README.md
  • QUICKSTART.md
  • agents/01-inbox-triager.md
  • agents/02-research-agent.md
  • agents/03-support-agent.md
  • agents/04-lead-qualifier.md
  • agents/05-content-repurposer.md
  • agents/06-code-reviewer.md
  • agents/07-data-analyst.md
  • agents/08-meeting-notes.md
  • agents/09-recruiting-screener.md
  • agents/10-sales-email-writer.md
  • agents/11-docs-qa.md
  • agents/12-bug-triager.md
  • agents/13-personal-assistant.md
  • agents/14-finance-categorizer.md
  • agents/15-orchestrator.md
  • CHANGELOG.md
  • code/ai-sdk-agent.ts
  • code/evals.demo.ts
  • code/evals.ts
  • code/openai-agents.ts
  • code/package.json
  • code/README.md
  • code/tsconfig.json
  • design-guide.md
  • examples/before-after.md

README preview

# Agent Builder

**Version 1.0.0. Verified October 15, 2026 against AI SDK 7 and OpenAI Agents SDK 0.20.**

Fifteen production-ready agent designs. Each one includes a system prompt, typed tool schemas, guardrails and eval cases. The pack also includes working TypeScript for the AI SDK and the OpenAI Agents SDK that we type-checked before release. It replaces the old "Agent System Prompts" pack: buyers of that pack get this one at the same download link.

## What's inside

| Path | What it is |
|---|---|
| `QUICKSTART.md` | From zero to a running agent with evals |
| `design-guide.md` | The five decisions that matter: scope, tools, prompt, untrusted input, evals |
| `agents/` | 15 agent designs (list below) |
| `code/` | Type-checked AI SDK and OpenAI Agents examples, plus an eval harness |
| `examples/before-after.md` | Three v1 prompts upgraded, with the reasons |
| `CHANGELOG.md` | Version history |

## The 15 agents

1. Inbox triager: drafts only, resists prompt injection
2. Research agent: citations limited to pages it actually fetched
3. Support reply agent: knowledge base first, terminal escalate tool
4. Lead qualifier: scoring rubric and structured output
5. Content repurposer: no invented statistics


+ Your logo here