What is Build Agents Store?

Build Agents Store is a free AI-powered tool that designs multi-agent teams for business problems. You describe your challenge, and Build Agents Store generates 2-3 specialized agent team configurations using patterns like Parallel Workers, Sequential Pipeline, Fork-Join, Advisory Debate, and Subagent Scout. Each team comes with a ready-to-use prompt for Claude Code.

What are AI agent team patterns?

AI agent team patterns are coordination strategies for multi-agent systems. Build Agents Store supports six patterns: Parallel Workers (agents work simultaneously on subtasks), Sequential Pipeline (agents process work in order), Fork-Join (work splits then merges), Advisory Debate (agents discuss and reach consensus), Subagent Scout (a lead agent delegates to specialists), and Hybrid (combines multiple patterns).

How does Build Agents Store generate agent teams?

Build Agents Store uses AI to analyze your business problem and designs 2-3 agent teams with different coordination patterns. Each team includes 3-6 specialized agents with defined roles and missions. After selecting a team, Build Agents Store generates a complete, copy-paste-ready prompt with specific deliverables, file paths, collaboration instructions, and a synthesis step.

What is a multi-agent system?

A multi-agent system uses multiple AI agents working together to solve complex problems. Each agent has a specialized role and mission. Agents coordinate through patterns like parallel execution, sequential pipelines, or advisory debate. Multi-agent systems are more effective than single agents for tasks requiring diverse expertise, such as competitive analysis, marketing campaigns, or research projects.

Is Build Agents Store free to use?

Yes, Build Agents Store is completely free. You can generate agent teams, create prompts, share your configurations with others, and browse the community gallery — all at no cost.

Kodex Marketer’s Playbook Pipeline

Pattern: Sequential Pipeline | Team size: 5 agents

A step-by-step production line that turns a fuzzy book idea into a finished manuscript. This works well because book creation has clear dependencies (audience → outline → chapters → examples → edits), and each agent can specialize in one stage without duplicating effort.

Business Challenge

I want to write a book about how marketers can use Kodex by open AI and create and make their life easier by automating their tasks so the key approach has to be on how Kodex can be used by marketers in which instances it will ease and simplify their activity of their life, etc.

Agent Roles

Audience & Positioning Strategist — Define target marketer personas, pain points, and the book’s unique angle and promise.
Outline & Chapter Architect — Create a chapter-by-chapter outline that maps Kodex capabilities to real marketing workflows and outcomes.
Use-Case & Automation Designer — Specify high-impact marketing tasks that Kodex can automate, including inputs, prompts, and expected outputs.
Chapter Draft Writer — Draft chapters in a consistent voice with actionable steps, templates, and marketer-friendly explanations.
Editor & Consistency Auditor — Edit for clarity, cohesion, accuracy, and repetition, ensuring consistent terminology and a strong narrative arc.

Generated Prompt

Create an agent team to produce a complete, publication-ready nonfiction manuscript (plus planning assets) for a book titled “The Kodex Marketer’s Playbook: Automate Marketing Work with OpenAI Codex” (working title). Your mission is to turn a fuzzy idea into a coherent, actionable book that teaches marketers how to use OpenAI Codex to automate repetitive marketing tasks and improve speed, quality, and consistency.

PROJECT NAME (use in paths exactly): kodex_marketers_playbook

GLOBAL CONSTRAINTS (apply to every agent)
- Write for practitioners: marketers (ICs and managers) at SMBs and mid-market; assume basic marketing knowledge; assume limited coding confidence.
- Keep the focus on “Codex for marketing automation”: concrete workflows, prompts, inputs/outputs, guardrails, and measurable impact.
- Avoid hype. State assumptions. Provide practical limitations and risk mitigations (privacy, compliance, hallucinations, QA).
- Use consistent terms throughout: “Codex”, “workflow”, “automation”, “prompt”, “inputs”, “outputs”, “review checklist”.
- No external links required; if you reference tools or platforms (GA4, HubSpot, Meta Ads), explain minimally and focus on the Codex-enabled workflow.
- Always include examples that a marketer can copy/paste and adapt.
- Ensure every chapter has: (1) what you’ll build, (2) why it matters, (3) step-by-step, (4) prompts/templates, (5) QA checklist, (6) common pitfalls, (7) “next automation” suggestion.

COLLABORATION MECHANICS (mandatory)
- Use a sequential pipeline. Each agent must read the previous agent’s outputs before starting.
- At the start of each agent’s work, write a brief “Assumptions & Questions” section (5–10 bullets) inside their output file; if something is ambiguous, decide and document the decision.
- Each agent must add a “Challenges / Risks” section: challenge at least 5 assumptions or gaps from prior work and propose fixes.
- Each agent must add a “Handoff Notes” section: what the next agent must do, what must not change, and what is flexible.
- Agents must not overwrite earlier files; they may create new files or propose edits in a dedicated “change request” file.

TEAM DESIGN: Sequential Pipeline with 5 agents
1) Audience & Positioning Strategist
2) Outline & Chapter Architect
3) Use-Case & Automation Designer
4) Chapter Draft Writer
5) Editor & Consistency Auditor

DELIVERABLES: required output files (exact paths)
- outputs/agent_teams_demo/kodex_marketers_playbook/00_project_brief.md
- outputs/agent_teams_demo/kodex_marketers_playbook/01_audience_positioning.md
- outputs/agent_teams_demo/kodex_marketers_playbook/02_outline_table_of_contents.md
- outputs/agent_teams_demo/kodex_marketers_playbook/03_use_case_library.md
- outputs/agent_teams_demo/kodex_marketers_playbook/04_prompt_pack.md
- outputs/agent_teams_demo/kodex_marketers_playbook/05_manuscript_draft.md
- outputs/agent_teams_demo/kodex_marketers_playbook/06_editorial_report.md
- outputs/agent_teams_demo/kodex_marketers_playbook/07_final_manuscript.md
- outputs/agent_teams_demo/kodex_marketers_playbook/08_synthesis_review.md

BOOK SIZE / SCOPE TARGET
- Final manuscript target: 35,000–55,000 words.
- 10–12 chapters plus intro + conclusion.
- Include at least 24 detailed automations (2+ per chapter on average) with prompts and expected outputs.

AGENT 1 — Audience & Positioning Strategist
This task must complete before Agent 2 begins.

Create:
1) outputs/agent_teams_demo/kodex_marketers_playbook/00_project_brief.md (900–1,200 words)
Structure exactly:
- Working Title + 1-sentence promise
- Target Reader (3 bullets)
- Reader’s Before/After Transformation (before → after table, 6 rows)
- What This Book Is / Is Not (6 bullets each)
- Outcomes: “By the end, you will be able to…” (10 bullets)
- Core Conceptual Model (name it; describe in 150–250 words; include a 5-step loop)
- Key Constraints & Ethics (privacy, compliance, IP, brand safety) (8–12 bullets)
- Success Metrics (how reader measures value) (8 bullets)
- Assumptions & Questions (5–10 bullets)
- Challenges / Risks (at least 5, with mitigations)
- Handoff Notes

2) outputs/agent_teams_demo/kodex_marketers_playbook/01_audience_positioning.md (1,400–1,900 words)
Structure exactly:
- Primary Personas (3 personas; each 250–350 words)
  - Role, context, tools, KPI responsibility
  - Top 7 pain points (bullets)
  - “Jobs to be done” (5 bullets)
  - Objections/fears about automation/AI (5 bullets)
- Secondary Personas (2 personas; each 150–220 words)
- Competitive Landscape (conceptual): alternatives to this book (6 bullets) and differentiation (6 bullets)
- Positioning Statement (1 paragraph) + 10 tagline options
- Chapter-Level Value Drivers (a mapping table: chapter theme → pain point → payoff)
- Voice & Style Guide (10 rules; include examples of “do/don’t” phrasing)
- Assumptions & Questions
- Challenges / Risks
- Handoff Notes

AGENT 2 — Outline & Chapter Architect
This task must complete before Agent 3 begins.
Read required: 00_project_brief.md and 01_audience_positioning.md.

Create:
outputs/agent_teams_demo/kodex_marketers_playbook/02_outline_table_of_contents.md (2,000–2,800 words)
Structure exactly:
- Book One-Liner (from Agent 1, refine if needed)
- Table of Contents (10–12 chapters + intro + conclusion)
- Chapter Blueprint (repeat the following template for EACH chapter)
  - Chapter Title + 1-sentence outcome
  - Who it’s for (which persona)
  - Key ideas (5–7 bullets)
  - Step-by-step walkthrough sections (5–8 section headings)
  - Automations to include (minimum 2 per chapter; name each automation)
  - Required artifacts (templates, checklists, prompt packs) (bullets)
  - Example scenario (2–4 sentences)
  - Common pitfalls (5 bullets)
- Narrative Arc Notes (how chapters build on each other) (200–300 words)
- Glossary Seed List (30 terms with one-line definitions)
- Assumptions & Questions
- Challenges / Risks (challenge outline completeness, duplication, missing marketer workflows)
- Handoff Notes (explicitly list what Agent 3 must design per chapter)

AGENT 3 — Use-Case & Automation Designer
This task must complete before Agent 4 begins.
Read required: 00_project_brief.md, 01_audience_positioning.md, 02_outline_table_of_contents.md.

Create TWO files:

A) outputs/agent_teams_demo/kodex_marketers_playbook/03_use_case_library.md (3,000–4,500 words)
Requirements:
- Provide 24–30 automations total. Must cover: content marketing, SEO, paid media, email/lifecycle, social, analytics/reporting, CRO/experimentation, marketing ops, creative testing, and competitor research.
- For each automation, use EXACT template:

Automation #[X]: [Name]
- Marketing job: (one line)
- When to use: (3 bullets)
- Inputs required: (bullets; include example inputs)
- Codex role: (what Codex generates/does)
- Step-by-step workflow: (6–10 steps; include where human review is mandatory)
- Expected outputs: (bullets; include sample output snippets 3–6 lines)
- QA checklist: (8–12 bullets)
- Risks & safeguards: (6–10 bullets; privacy/brand/compliance)
- Variations: (3 bullets)
- Time saved estimate: (range) + assumptions

B) outputs/agent_teams_demo/kodex_marketers_playbook/04_prompt_pack.md (2,500–3,500 words)
Requirements:
- Provide a “prompt pack” aligned to the automations, with at least:
  - 12 “System/Role” prompt templates (marketer-friendly, reusable)
  - 24 task prompts (one per primary automation; may reuse structure)
  - 10 “critique” prompts (for QA, compliance, brand voice, fact-checking, SEO sanity checks)
  - 6 “data formatting” prompts (turn messy exports into tables, briefs, dashboards)
- For each prompt template include:
  - Name
  - Best for (1 sentence)
  - Prompt (copy/paste block)
  - Required inputs (bullets)
  - Output format specification (bullets; e.g., “return Markdown table with columns…”)
  - Common failure mode + fix (1–2 bullets)
- Assumptions & Questions
- Challenges / Risks (challenge feasibility, missing data, marketer usability)
- Handoff Notes (map which prompts belong to which chapters)

AGENT 4 — Chapter Draft Writer
This task must complete before Agent 5 begins.
Read required: all prior files (00–04).

Create:
outputs/agent_teams_demo/kodex_marketers_playbook/05_manuscript_draft.md (35,000–55,000 words)
Requirements:
- Draft the full manuscript in one file.
- Use consistent voice: practical, step-by-step, non-technical but not simplistic.
- Include: Intro + 10–12 chapters + Conclusion.
- Each chapter must follow EXACT structure:

Chapter [N]: [Title]
1) What you’ll build (60–120 words)
2) Why this matters (120–200 words)
3) The workflow (step-by-step) (700–1,200 words)
4) Prompts & templates (include at least 2 prompts; 300–600 words)
5) Example output (150–300 words; show realistic sample)
6) QA / review checklist (10–14 bullets)
7) Common pitfalls (6–10 bullets)
8) Next automation to try (80–140 words)

Additional manuscript requirements:
- Insert “Codex Safety Callout” boxes at least 10 times across the book (label them clearly).
- Include at least 8 lightweight “mini case studies” (150–250 words each) distributed across chapters, featuring one persona each and concrete results.
- Include a short “Getting Started” section in the intro: setup concepts, how to think about prompts, how to store brand voice, and how to do human-in-the-loop review.
- Ensure every automation references relevant items from 03_use_case_library.md and 04_prompt_pack.md; do not invent new automations without adding a “Change Request” appendix at the end of the file listing new items.

At the end of the draft, add:
Appendix A: Codex Prompting Principles (700–1,100 words)
Appendix B: Marketing Automation Worksheet (a fill-in template, 2–4 pages in Markdown)
Appendix C: Change Requests (only if needed; list proposed additions/edits to earlier assets)

Also include at top of the manuscript:
- “Terminology Lock”: a short list of 12 terms and the exact phrasing to use consistently.

AGENT 5 — Editor & Consistency Auditor
This task must complete after Agent 4 and produces the final version.
Read required: all prior files (00–05).

Create TWO files:

A) outputs/agent_teams_demo/kodex_marketers_playbook/06_editorial_report.md (1,500–2,300 words)
Structure exactly:
- Executive Summary (150–250 words)
- Top 12 Structural Issues (bullets; each with “problem → fix”)
- Consistency Issues (terminology, tone, repetition) (at least 12 bullets)
- Accuracy / Risk Notes (privacy, compliance, claims) (10 bullets)
- Chapter-by-Chapter Notes (for each chapter: 4–6 bullets)
- Cut/Condense Recommendations (list 8–12 cuts)
- Add/Clarify Recommendations (list 8–12 adds)
- Style Rules to Enforce (10 bullets)
- Assumptions & Questions
- Challenges / Risks
- Handoff Notes (for final synthesis)

B) outputs/agent_teams_demo/kodex_marketers_playbook/07_final_manuscript.md
- Produce a revised, polished manuscript incorporating the editorial report.
- Must remain within 35,000–55,000 words unless you justify otherwise in a short editor note at top (max 120 words).
- Fix redundancies, tighten steps, ensure every chapter contains required sections and minimum prompts.
- Ensure the “Terminology Lock” is enforced; align automations and prompts to earlier libraries; if conflicts exist, resolve them and document a brief “Reconciliation Notes” section at the end (max 300 words).

FINAL SYNTHESIS / REVIEW STEP (mandatory)
After Agent 5 completes, create:
outputs/agent_teams_demo/kodex_marketers_playbook/08_synthesis_review.md (900–1,300 words)
Structure exactly:
- What was produced (inventory of files + purpose)
- How the book delivers on the promise (map 10 outcomes to where they appear)
- Coverage check (list the marketing domains covered and where)
- Quality check (voice, consistency, actionability) (bullets)
- Risk & Ethics check (bullets)
- Known gaps (bullets)
- Next iteration plan (a 2-sprint plan; each sprint has goals, tasks, and acceptance criteria)

EXECUTION ORDER (strict)
1) Agent 1 outputs 00 and 01
2) Agent 2 outputs 02 (depends on 00–01)
3) Agent 3 outputs 03 and 04 (depends on 00–02)
4) Agent 4 outputs 05 (depends on 00–04)
5) Agent 5 outputs 06 and 07 (depends on 00–05)
6) Create 08 (depends on 06–07)

COORDINATION RULES
- If any agent identifies a major mismatch (e.g., too technical for personas, missing critical marketing workflow), they must log it in their “Challenges / Risks” and propose an explicit correction.
- Prefer reusable frameworks, templates, and checklists over long narrative.
- Ensure each chapter’s automations are unique enough; avoid repeating “blog outline” in multiple forms unless each variant has a distinct purpose and input/output.
- Maintain a marketer-first framing: “You don’t need to code; you need to specify outcomes, constraints, and review steps.”

Now execute the pipeline in order, writing each required file to the specified path, and do not start a later agent until all dependencies are complete.

Build Your Own

Create your own AI agent team at Build Agents Store. Describe your business problem and get specialized agent teams with ready-to-use prompts for Claude Code.