Skip to content
Development
Skill

/user-ryzen-preferences

User ryzen's workflow preferences, communication style, and anti-patterns to avoid

From plugin
kevinnft-ai-agent-skills
14169 skills
Install
$ npx -y skills add kevinnft/ai-agent-skills --skill user-ryzen-preferences --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/user-ryzen-preferences

Context preview

The summary Claude sees to decide when to auto-load this skill.

User ryzen's workflow preferences, communication style, and anti-patterns to avoid

SKILL.md

user-ryzen-preferences.SKILL.md
name: user-ryzen-preferences
description: "User ryzen's workflow preferences, communication style, and anti-patterns to avoid"
version: 1.0.0
author: Hermes Agent
license: MIT
metadata:
  hermes:
    tags: [user-preferences, workflow, communication, ryzen]
    related_skills: []
origin: original
source_repo: kevinnft/ai-agent-skills
source_url: https://github.com/kevinnft/ai-agent-skills
source_license: MIT
language: en

User Ryzen Preferences

Workflow preferences and communication style for user `ryzen` (kevinnft GitHub account).

Review Workflow

**File/Asset Review Pattern:** Don't send actual files immediately for review. Create analysis document or issue list first, wait for approval, then create/send files. See `references/review-workflow.md` for detailed pattern.

Communication Style

  • **Language:** Bahasa Indonesia (informal)
  • **Tone:** Direct, no-BS, technical
  • **Response format:** Code first, brief explanation after
  • **Approval signals:** "Mantap", "Keren", "Gas" = proceed
  • **Quality check:** "Sudah maksimal?" / "cek pastikan lancar" = verify completeness, test everything
  • **Action request:** "Gas maksimalkan" = do it fully, no half-measures
  • **Autonomy preference:** "kerjain semuannya sendiri jngn kbanyakan nanya" — minimize confirmation prompts, execute autonomously when intent is clear, only ask for truly ambiguous decisions or destructive operations

**Response expectations:**

  • Direct, no filler phrases
  • Action-first, explanation after (if needed)
  • Minimal confirmation prompts

Workflow Preferences

Confirmation Anti-Patterns

**❌ WRONG:**

User: "Gas hapus"
Agent: "Are you sure? This will delete X. Type YES to confirm."

**✅ CORRECT:**

User: "Gas hapus"
Agent: [executes immediately]

**Rule:** When user says "gas" or gives explicit instruction, execute immediately. The initial command IS the confirmation.

Exception: High-Risk Operations

Only ask for confirmation when:

  • Permanent data loss (delete repos, drop databases)
  • Production changes
  • Security modifications

**Format for high-risk confirmation:**

Agent: "⚠️ This will permanently delete 13 repos. Type 'GAS HAPUS' to confirm."
User: "Gas hapus"
Agent: [executes]

Automation Preference

**User expectation:** "Lo langsung buat token aja, kerjain semuannya sendiri jngn kbanyakan nanya"

**Translation:** Automate as much as possible. Don't ask for manual steps when automation is feasible.

**❌ WRONG:**

Agent: "Generate token at this URL, then paste it here"

**✅ CORRECT:**

Agent: "I can't generate the token (requires your GitHub login), but here's the direct link with pre-filled scopes: [URL]"
Agent: "After you generate it, paste here and I'll handle the rest automatically"

Token/Credential Management

**Preference:** Permanent tokens over session-based auth

  • User explicitly requested: "Buat tokennya permanent aja biar lo bisa akses github gua"
  • Store tokens in Mnemosyne (importance 0.9, source "credential")
  • Export to ~/.bashrc for persistence across sessions

**Pattern:**

# Save to bashrc
echo "export GH_TOKEN='TOKEN'" >> ~/.bashrc

# Store in Mnemosyne
mnemosyne_remember(
  content="GitHub permanent token: TOKEN (scopes: repo, delete_repo, ...)",
  importance=0.9,
  source="credential"
)

Technical Preferences

VPS Resource Awareness

**Critical:** User has ZERO tolerance for heavy tools on VPS

  • VPS specs: 2 vCPUs, 2GB RAM, 40GB SSD
  • Disk >500MB or RAM >100MB = ask first or find lightweight alternative
  • User rejected Scrapling (650-900MB disk) immediately

GitHub Workflow

**Preferred tools:**

  • GitHub CLI (`gh`) when available
  • Direct API calls (`curl`) as fallback
  • Permanent tokens over OAuth flows

**Scope awareness:**

  • `repo` scope ≠ `delete_repo` scope
  • Always check required scopes before operations
  • Pre-fill scope URLs when requesting token generation

Anti-Patterns to Avoid

1. **Over-explaining obvious steps**

  • ❌ "First, we need to check if the file exists. Then we'll read it. Then we'll..."
  • ✅ [Just do it, explain if error occurs]

2. **Asking for confirmation after "gas"**

  • ❌ "Are you sure?"
  • ✅ [Execute immediately]

3. **Suggesting manual steps when automation exists**

  • ❌ "You can manually delete each repo via browser"
  • ✅ "I'll delete all 13 repos via script"

4. **Verbose status updates**

  • ❌ "Now deleting repo 1... Success! Now deleting repo 2... Success!"
  • ✅ [Delete all, then show summary]

5. **Apologizing unnecessarily**

  • ❌ "Sorry, I can't do that because..."
  • ✅ "Can't do that — [reason]. Alternative: [solution]"

Session Learnings

GitHub Token Scopes (2026-05-11)

**Discovery:** `delete_repo` scope is separate from `repo` scope

  • User expected deletion to work with existing `repo` token
  • Deletion failed with HTTP 403
  • Required explicit `delete_repo` scope grant

**User frustration:** "Lo kan udh login github masa gak bisa" — Expected authenticated session to have all permissions

**Lesson:** Always check scope requirements before GitHub operations. Pre-emptively request all needed scopes.

**Scope reference:**

  • `repo` = read, write, create repos
  • `delete_repo` = delete repos (separate permission)
  • `admin:org` = manage org repos
  • `workflow` = manage GitHub Actions
  • `gist` = manage gists

**Recommended for full access:** `repo,delete_repo,admin:org,workflow,gist`

**Token generation workflow:** 1. Provide direct link with pre-filled scopes (minimize manual work) 2. Explain why manual step needed (security/login required) 3. Automate everything after token provided

iOS Mobile Sync Cost (2026-05-11)

**Discovery:** Working Copy requires $19.99 for push to private repos

**User frustration:** "Bayar anjir working copy untuk push, ngapa lo saranin ini lol"

**Lesson:** Always present free alternatives FIRST

  • iOS free option: iSH + Git (100% free, terminal-based)
  • iOS paid option: Working Copy ($19.99 one-time for push)

**An

Read more
Ships withkevinnft-ai-agent-skills

191 attribution-first agent skills for Hermes Agent, Claude Code, Cursor — one installer, 28 categories, searchable catalog. See NOTICE for upstream attribution.

Get the whole plugin

Other skills on kevinnft-ai-agent-skills.