AGENT SKILL

ai-team-blueprint

Guide a solo operator or small business from "I want an AI team" to a ready-to-run v1 blueprint for a team of 3 to 5 Claude agents. Asks five sharp questions, recommends one starting use case, then delivers the team structure, a production prompt per agent, the handoff workflow, a lean tool stack, a QA plan and a 7-day launch plan ending in START HERE FIRST. Use whenever someone wants to set up their first AI agents or multi-agent team, asks which agents they need, how to split work between agents, how to get Claude to take a recurring chunk of their work off their plate, or how to use AI agents for content, lead generation, support, research or operations, even if they never say "blueprint" or "team". Not for writing agent code against the Claude API or Agent SDK (use claude-api), building a single skill (use draft-skill), running agents on a task right now, or picking one AI tool for a one-off task.

License 许可
MIT
最近更新
2026-11-14
Stars
★ 0
Forks
0

01 前言 Front Matter

name
ai-team-blueprint
version
1.0.0
description
Guide a solo operator or small business from "I want an AI team" to a ready-to-run v1 blueprint for a team of 3 to 5 Claude agents. Asks five sharp questions, recommends one starting use case, then delivers the team structure, a production prompt per agent, the handoff workflow, a lean tool stack, a QA plan and a 7-day launch plan ending in START HERE FIRST. Use whenever someone wants to set up their first AI agents or multi-agent team, asks which agents they need, how to split work between agents, how to get Claude to take a recurring chunk of their work off their plate, or how to use AI agents for content, lead generation, support, research or operations, even if they never say "blueprint" or "team". Not for writing agent code against the Claude API or Agent SDK (use claude-api), building a single skill (use draft-skill), running agents on a task right now, or picking one AI tool for a one-off task.

02 资源文件 4

03 SKILL.md 正文

ai-team-blueprint

Turn a vague wish for "an AI team" into a small, real system one person can run. The output is a single blueprint document. The work is mostly judgement: picking the narrowest use case that pays off, and keeping the team small enough to actually launch.

Rules that apply throughout

  • Narrow beats impressive. Recommend 3 agents unless the use case clearly needs a 4th or 5th. Every extra agent adds a handoff that can break, and a beginner debugging five agents gives up.
  • Real, not a demo. Every agent must produce an output a human would otherwise make by hand. If an agent only "coordinates" or "thinks", cut it or merge it.
  • Evidence before big decisions. Base the use-case pick on the user's answers, and say which answer drove it. If the answers conflict, ask instead of guessing.
  • Human approval gates. Every agent prompt and the workflow must stop for the user's approval before publishing, sending, deleting, spending money, or any irreversible change. These gates are the main safety mechanism for a beginner's first system, so never drop one to make the flow smoother.
  • No secrets in chat. Never ask for passwords, API keys or tokens. When a tool needs credentials, say where the user enters them (the tool's own settings or an environment secret), not that they should paste them.
  • Model-agnostic. Recommend "the current most capable Claude model" for reasoning-heavy agents and a faster, cheaper Claude model for high-volume simple steps. Don't hardcode a version name, because the blueprint will outlive it.

Workflow

1. Ask five sharp questions, once

Ask them in one message, numbered, and skip any the user already answered:

  1. What outcome do you want in the next 90 days? (Revenue, hours saved, output volume.)
  2. What are your skills? Can you edit a spreadsheet, use no-code automation, or write a little code?
  3. How many hours a week can you give this, both to set it up and to run it?
  4. Do you already have a use case in mind?
  5. What is the first concrete task you want off your plate?

Infer from the conversation where you can, for example from connected tools or the user's business. If the user wants to skip the questions, state your assumptions and go on.

2. Recommend one use case

Score the realistic options against the answers using references/use-cases.md, and show a short comparison of the top three. Recommend one, give a one-line reason tied to the user's answers, and ask for a yes before building. The whole blueprint hangs on this choice, so it is the one point worth pausing for.

3. Build the blueprint

Follow references/blueprint-template.md exactly; where it and this file disagree, this file wins. It fixes the section order, the per-agent spec, the prompt format and the 7-day plan shape. Write each agent prompt so it can be pasted straight into a Claude Project or system prompt.

Keep agents from overlapping: each owns one artifact, and the workflow section names who hands what to whom. If two agents could both produce the same thing, merge them.

4. Deliver

The blueprint is long-lived material the user will edit and come back to, so put it in a document or page when the environment offers one, and keep chat for a short summary. Otherwise give it as Markdown. End with START HERE FIRST: the three actions for today, each under 30 minutes.

Handoffs

  • To code: if the user wants the team built in code with the Claude API or Agent SDK, offer one line handing over to claude-api. Pass the team-structure table, the agent prompts and the workflow section as they are. If claude-api isn't installed, say so and stop at the blueprint. Don't write the code here.
  • Doing, not designing: if the user wants agents run on a task now, skip the blueprint and just do the task.

Check before handing over

  • 3 to 5 agents, each with a distinct output and named handoffs.
  • Every agent prompt has an escalation section with the approval gates.
  • The tool stack has no tool the 7-day plan doesn't use.
  • The QA plan has at least one metric that shows whether the team actually saves time or makes money.
  • No hardcoded model version, and no request for secrets.

04 同仓库更多 Skills

arm-pz
arm-pz/caffeinity-skills

tune-and-ship

Takes one published skill at a time from as-is to tuned, fact-checked, released and live, then moves to the next one in the catalog. Each run tunes the skill with tune-skill, checks every version number, flag, price and command it states against the primary source, re-reads its siblings' real SKILL.md files, polishes the listing with findable-polish, writes the version and release notes, opens one PR, and after the merge confirms the site page and marketplace copy are live. Use whenever the user wants to fine-tune their other skills, work through the catalog, give a published skill the full treatment, fact-check a skill's claims, or ship a tune instead of drafting it, even if they never say "ship". Not for a tune with no release (use tune-skill), drafts with nothing pushed (use tune-and-polish), a new skill (use skill-launch), regressions from the nightly health check (use caffeinity-routing's Descale brew), or a whole collection (use skill-estate).

★ 0
arm-pz
arm-pz/caffeinity-skills

skill-estate

Takes a whole collection of agent skill repos from scattered to published, consistent and self-maintaining, in gated phases: inventory every repo, reconcile site, marketplace and README listings, test and tune the family, close workflow gaps, wire publishing so new skill repos go live on their own, bundle a plugin with agents, add trigger tests and directory listings, and leave a maintenance loop and a map. Each phase hands off to the skill that owns it and ends in PRs. Use when someone's skills site and GitHub disagree, wants all their skill repos audited or cleaned up, wants new skills to publish themselves, or wants that cleanup repeated on another org, even if they never say estate. Not for one skill (use tune-skill), one listing (use findable), the site pipeline alone (use publish-pipeline), or weekly upkeep (use caffeinity-routing's Descale brew).

★ 0
arm-pz
arm-pz/caffeinity-skills

paper-trail

Drafts the policy pages a small product needs before it takes money or data (privacy policy, terms of service, refund policy, cookie notice) from a fact sheet about what the product actually collects, stores, shares and charges, then checks the drafts for placeholders, missing sections and facts nobody supplied. Use when someone needs a privacy policy, terms, refund or cookie page for a site, app, SaaS, newsletter or digital product, asks what legal pages they need before launch, or wants existing policy pages checked against what the product really does, even if they never say "legal". Not legal advice and not for contracts, NDAs or employment documents (use a lawyer), and not for the launch plan itself (use solo-income-blueprint) or the site build (use site-prompt).

★ 0
arm-pz
arm-pz/caffeinity-skills

course-builder

Designs a course for a group of learners (a class, cohort, workshop series, bootcamp or internal training): measurable outcomes, a module sequence with prerequisites, activities and assessments aligned to each outcome, a time budget that fits the hours available, and a rubric, then checks the plan with a script so no outcome goes untaught or unassessed. Use when someone is teaching others and asks for a curriculum, syllabus, course outline, module plan, cohort or workshop structure, or wants an existing course checked for gaps and overload, even if they never say "curriculum". Not for one person's own learning plan (use study-system), for being taught the topic (use topic-primer), or for marketing the course (use sales-copy).

★ 0
arm-pz
arm-pz/caffeinity-skills

wordsmith

Write, draft, rewrite, edit, proofread, polish and improve text while preserving the author's intended meaning, facts and voice. Use when the user wants clearer, shorter or better-toned writing; grammar, structure or flow fixes; emails, website or marketing copy, headlines, outlines or summaries; a different reading level; an accessibility, inclusivity or accuracy review; or content prepared for publication, even if they never say "edit". Also handles /wordsmith commands such as draft, critique, audit, tighten, tone and final. Not for resumes, CVs or LinkedIn profiles (use recruiter-lens if installed), not for code or code comments, and not for building .docx or .pdf files (write the text here, then use the file skill).

★ 0
arm-pz
arm-pz/caffeinity-skills

web-preflight

Use when shipping, deploying, publishing, launching, reviewing, or pre-flighting a website or API and you want a lightweight check for accidental exposure of secrets, source or configuration files, admin and debug surfaces, unsafe redirects, missing security controls, and misleading 200 responses. Trigger phrases include "before we ship", "security check", "preflight", "audit the site", "check the production URL", "is my .env exposed", and "why does /.env return 200", including when access-log or analytics evidence is supplied. Checks only sites the user owns or is authorized to assess, with bounded GET and HEAD requests and a fixed compact path set; distinguishes genuine resources from SPA catch-all and blanket WAF responses. Not for penetration testing, exploitation, credential testing, broad port scanning, or application-code review; use a dedicated authorized security-testing or code-review workflow instead.

★ 0