agentleFS
minhnv0807minhnv0807/fullstack-mkt-skills/AGENTS.md

fullstack-mkt-skills

AGENTS.md644 starsChanged 6 months agoMITView on GitHub

Huong dan cho bat ky AI agent nao (khong chi Claude) lam viec voi repo nay. Repo nay chua Agent Skills theo Agent Skills Specification — cai dat duoc vao .agents/skills/ (hoac .claude/skills/ voi Claude Code). Repo cung hoat dong nhu Claude Code Plugin Marketplace — cai dat bang /plugin install minhnv0807/ai-business-skills. Moi SKILL.md phai co: product-marketing-context la skill goc — moi skill khac doc truoc khi hoat dong. File .agents/product-marketing-context.md chua: - Product overview - Target audience - Personas…

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this AGENTS.md into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read both context files (skill checks existence)

    line 143
  2. Distinctive:

    If only A → Mode A

    line 144
  3. Distinctive:

    If only B → Mode B

    line 145
  4. Distinctive:

    If both → Skill ASKS user

    line 146
  5. Distinctive:

    If neither → Skill SUGGESTS creating context first

    line 147

≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md — Fullstack Marketing Skills

Huong dan cho bat ky AI agent nao (khong chi Claude) lam viec voi repo nay.

Tong quan

Repo nay chua Agent Skills theo Agent Skills Specification — cai dat duoc vao .agents/skills/ (hoac .claude/skills/ voi Claude Code).

Repo cung hoat dong nhu Claude Code Plugin Marketplace — cai dat bang /plugin install minhnv0807/ai-business-skills.

Cau truc repo

.
├── .claude-plugin/
│   └── marketplace.json         # Claude Code plugin manifest
├── .github/
│   ├── ISSUE_TEMPLATE/           # Issue templates
│   └── PULL_REQUEST_TEMPLATE/    # PR templates
├── skills/
│   └── skill-name/
│       ├── SKILL.md              # Skill definition (required)
│       ├── references/           # Supporting docs (optional)
│       ├── scripts/              # Helper scripts (optional)
│       └── evals/evals.json      # Trigger / routing / guardrail tests (xem docs/evals.md)
├── references/                   # Shared references across skills
├── knowledge/                    # Foundation thinking (upload as AI project context)
│   └── trien-khai/               # Implementation knowledge by role
├── workflows/                    # Multi-skill workflows
├── agents/                       # Agent personas/roles
├── examples/                     # Sample outputs
├── AGENTS.md                     # This file
├── CLAUDE.md                     # Claude-specific instructions
├── VERSIONS.md                   # Version tracking (bang sinh tu metadata.version)
├── NOTICE.md                     # Nguon y tuong ben thu ba + giay phep
├── scripts/check-repo.mjs        # Kiem tra nhat quan repo (evals, denylist, llms, manifest, versions, counts)
├── CONTRIBUTING.md               # Contribution guide
├── validate-skills.sh            # Spec validator (bash)
└── validate-skills.ps1           # Spec validator (PowerShell)

Yeu cau skill

Moi SKILL.md phai co:

1. YAML Frontmatter

---
name: skill-name                    # Required: 1-64 chars, lowercase + hyphens, match directory
description: "Mo ta cu the..."       # Required: 1-1024 chars, co trigger phrases
metadata:
  version: 1.0.0                    # Optional: semver
  category: strategy | content | performance | operations
license: MIT                        # Optional: default MIT
triggers:                           # Optional: keyword list for auto-activation
  - "trigger 1"
  - "trigger 2"
related:                            # Optional: skill cross-references
  - skill-1
  - skill-2
---

2. Noi dung

  • Duoi 500 dong — chi tiet chuyen vao references/
  • Ngon ngu ro rang, active voice
  • Cau truc H2 va H3
  • Co "Thu thap thong tin" section (toi da 4 cau)
  • Co output template co cau truc
  • Co quality checklist

Foundation skill

product-marketing-context la skill goc — moi skill khac doc truoc khi hoat dong.

File .agents/product-marketing-context.md chua:

  • Product overview
  • Target audience
  • Personas
  • Pain points
  • Competition
  • Differentiation
  • Objections
  • Customer language
  • Brand voice
  • Proof points
  • Goals

Skill khac kiem tra file nay truoc — neu co, lay thong tin san; neu khong, de xuat tao.

Pattern variants (Skill 20 + Skill 22)

Skills su dung pattern variants:

  • 20-brief-client-intake — 20 variants theo nganh hang
  • 22-personal-brand-context — 3 variants theo nhom audience (founder/coach/creator)

Cau truc:

modules/personal-branding/vi/22-personal-brand-context/
├── SKILL.md          ← Entrypoint + router
├── README.md         ← Variant guide (decision tree)
└── variants/
    ├── 01-founder.md
    ├── 02-coach.md
    └── 03-creator.md

User chay skill chinh, skill load README → user chon variant → skill load variant template.

Pattern variants — Skill 22 Global (NEW)

Skill 22-personal-brand-context-global uses 4 REGION variants (US/EU/SEA/LATAM) instead of 3 audience variants like VN.

Cau truc:

modules/personal-branding/en/22-personal-brand-context-global/
├── SKILL.md          ← Entrypoint + router
├── README.md         ← Region variant guide
└── variants/
    ├── 01-us.md      ← Covers founder + coach + creator (US)
    ├── 02-eu.md      ← Covers founder + coach + creator (EU)
    ├── 03-sea.md     ← Covers founder + coach + creator (SEA)
    └── 04-latam.md   ← Covers founder + coach + creator (LATAM)

Difference from VN skill 22:

  • VN: 3 audience variants (founder/coach/creator separately)
  • Global: 4 region variants (each contains 3 audience inside)
  • Reason: persona universal but currency/platforms/regulations differ per region

Pattern Mode-Switching (Skills 04, 05)

Skill 04, 05 dung pattern context-aware mode:

  • Doc 1 hoac nhieu file .agents/*.md
  • Auto-detect mode dua tren context file ton tai
  • Output adapt theo mode

Pattern dieu kien:

  1. Read both context files (skill checks existence)
  2. If only A → Mode A
  3. If only B → Mode B
  4. If both → Skill ASKS user
  5. If neither → Skill SUGGESTS creating context first

Pattern: Cluster Auto-Detect Mode (v2.5.0+)

Pattern used in v2.5.0 to handle 2 clusters (VN + Global) in same agents:

Agent reads `.agents/`:
- `product-marketing-context.md` → MODE VN
- `product-marketing-context-global.md` → MODE GLOBAL
- Both → ASK 1 question
- None → SUGGEST creating context

Same pattern for personal brand:

  • personal-brand-context.md → MODE VN PB
  • personal-brand-context-global.md → MODE GLOBAL PB

This pattern allows ONE agent to serve BOTH clusters without code duplication.

Used by: 6 agents (mkt-strategist, content-producer, performance-analyst, channel-operator, design-producer, personal-brand-builder).

Cai dat

Claude Code (khuyen dung)

# Plugin marketplace
/plugin install minhnv0807/ai-business-skills

# Hoac clone + install script
git clone https://github.com/minhnv0807/ai-business-skills.git
cd ai-business-skills
./install.sh --global

Agent khac (ChatGPT, Gemini, Cursor, v.v.)

  • Copy cac file .md lam Custom Instructions / System Prompt
  • Moi file la 1 prompt doc lap — khong can cai dat dac biet

Validation

Truoc khi commit, chay:

./validate-skills.sh

Script kiem tra:

  • Frontmatter dung spec (name, description, etc.)
  • Name khop voi directory
  • Duoi 500 dong
  • Co trigger phrases trong description

Sau do chay kiem tra nhat quan repo (Node 18+, khong can cai package):

node scripts/check-repo.mjs              # tat ca check
node scripts/check-repo.mjs --only=evals # mot check: evals | denylist | llms | manifest | versions | counts
  • evals (ISSUE) — moi evals/evals.json parse duoc, skill_name = ten thu muc, >=4 eval, moi eval co prompt, expected_output, >=2 assertions; eval should_trigger: false phai co routes_to tro toi skill co that. Moi file phai co it nhat 1 routing negative, 1 eval kind: guardrail, 1 eval kind: context; skill VN can it nhat 1 prompt co dau va 1 prompt khong dau (chi tiet: docs/evals.md).
  • denylist (ISSUE) — chan cac claim da bi go: ND 13/2023 (da thay the), ND 147/2024 lam can cu gan nhan AI, Luat Quang cao 2012 khong kem 75/2025, nguong similarity cua vendor, package @meta/ads-cli. Dong co cum "da thay the / bi thay the / thay the boi / superseded / deprecated / het hieu luc" duoc bo qua (ngu canh lich su); chu "thay the" don le khong du.
  • llms (ISSUE) — moi link trong llms.txt tro toi file co that.
  • manifest (ISSUE / WARN) — skill liet ke trong marketplace.json phai ton tai; moi version trong marketplace.json = plugin.json; skill chua liet ke la WARN.
  • versions (WARN) — bang VERSIONS.md khop metadata.version.
  • counts (WARN) — cum "NNN skills" trong README / README.vi / manifest khop so thu muc skill thuc te.

CI (.github/workflows/validate.yml) chay ca hai. Bat ky ISSUE nao lam fail build; WARN chi bao. Cach viet va cham evals: docs/evals.md.

Quy tac van hanh chung (v3.8.0)

Moi skill tro toi 8 quy tac: skills/vi/references/quy-tac-van-hanh.md (VN) / skills/en/references/operating-rules-global.md (Global). Phap ly VN chi dan ve mot hub: skills/vi/references/phap-ly-marketing-vn.md — khong ghi muc phat, thue suat, nguong hay % phi vao skill.

Versioning

  • Major (X.0.0): Breaking change (doi frontmatter, doi folder structure)
  • Minor (1.X.0): Them skill moi hoac them section quan trong
  • Patch (1.0.X): Sua loi, cap nhat so lieu, chinh syntax

Xem VERSIONS.md de theo doi thay doi.

Git & Contribution

Branch naming

  • feature/skill-name — skill moi
  • fix/skill-name — sua skill hien co
  • docs/... — cap nhat tai lieu

Commit format (Conventional Commits)

feat: add X skill
fix: correct benchmark in Y skill
docs: update README installation
refactor: restructure Z skill

PR process

  1. Fork repo
  2. Tao branch tu master
  3. Chay validate-skills.sh va node scripts/check-repo.mjs truoc khi push
  4. Tao PR voi mo ta ro rang
  5. Chi ra anh huong (skill nao bi affected, breaking change?)

Agent guidelines

Khi dung repo nay

  1. Doc product-marketing-context truoc — neu co file .agents/product-marketing-context.md, doc ngay truoc khi lam viec
  2. Doc skill tuong ung — khi nguoi dung yeu cau cong viec marketing, chon skill phu hop tu skills/
  3. Kiem tra version — doc VERSIONS.md 1 lan/session de biet ban cap nhat moi

Khi update skill

  1. Tang version trong metadata.version
  2. Cap nhat VERSIONS.md voi ngay + thay doi
  3. Giu frontmatter dung spec
  4. Them / cap nhat evals/evals.json khi doi routing hoac guardrail
  5. Chay validator + node scripts/check-repo.mjs truoc khi commit

Tuong thich

PlatformHo troCach dung
Claude CodeFull/plugin install hoac install.sh --global
Claude ProFullUpload vao Project Knowledge
ChatGPTPartialUpload .md lam Custom GPT config
GeminiPartialSystem prompt / context
CopilotPartial.github/copilot-instructions.md
CursorPartial.cursorrules hoac MDC files
WindsurfPartialGlobal rules
Agent khacPartialDoc .md nhu plain text context

Resources

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.