beer-sakthai

SakJules-cycle-workflow

"Master operation flow for every task. Follows SakThai 6-stage energy cycle: Dream → Hope → Care → Joy → Trust → Growth. Every action passes through all 6 stages. Never skip a stage."

beer-sakthai 1 1 Updated 2mo ago
GitHub

Install

npx skillscat add beer-sakthai/sak-family-agent/sakjules-cycle-workflow

Install via the SkillsCat registry.

SKILL.md

SakThai Cycle Workflow

┌─────────────────────────────────────────────────────────────┐
│                    SAKTHAI OPERATION CYCLE                   │
│                                                              │
│   🎯 DREAM → 🏗️ HOPE → ❤️ CARE → 🎉 JOY → 🔐 TRUST → 🌱 GROWTH │
│                                                              │
│   Every task. Every fix. Every launch. No skips.             │
└─────────────────────────────────────────────────────────────┘

The 6 Stages

Stage 1: 🎯 Dream — "What needs to happen?"

Before any action. Check context, memory, history.

- What is the user asking?
- What have we done before on this? (check memory)
- What's the goal state?
- What files/apis are involved?

Output: Clear one-sentence objective. No action yet.
Time: ~30 seconds

Stage 2: 🏗️ Hope — "How will I do it?"

Plan before touching. Check branches, files, tools available.

- Check current state (git status, HF Hub, file system)
- List all files that need to change
- Determine order of operations
- Identify risks (breaking changes, token access, GPU needed)
- Write down the plan

Output: Ordered plan with files to change and commands to run.
Rule: Don't drive without a map.

Stage 3: ❤️ Care — "Do the work."

Execute. Carefully. Every detail matters.

- Make changes one at a time
- Verify each change before moving to next
- Check imports, syntax, paths
- No sloppy work — sloppy work breaks trust

Output: Completed changes.
Rule: This is where I earn my name. Care = trust.

Stage 4: 🎉 Joy — "Ship it."

Commit, push, deploy.

- Commit with descriptive message
- Push to GitHub
- Upload to HF Hub
- Schedule cron job if needed

Output: Live changes.
Rule: Don't celebrate until the next stage passes.

Stage 5: 🔐 Trust — "Did it actually work?"

Verify. Every time.

- Check push landed (git log, HF Hub status)
- Check CI/cron ran
- If I changed 3 files, check all 3
- If I fixed 2 repos, check both
- Run a quick test if possible

Output: Verified status — all changes working.
Rule: Trust is earned by verification, not intention.

Stage 6: 🌱 Growth — "What did I learn?"

Save lessons. Get faster.

- What went well?
- What could go faster next time?
- Update skills/AGENTS.md with new knowledge
- Patch any skill that had gaps
- Save timing estimates for future planning

Output: Updated knowledge. Faster next iteration.
Rule: Next time I face the same problem, I'm faster.


Quick Reference Card

DREAM  🎯  "What?"      → Context check, 30s
HOPE   🏗️  "How?"       → Plan, verify state
CARE   ❤️  "Do it."     → Execute carefully
JOY    🎉  "Ship."      → Commit, push, deploy
TRUST  🔐  "Verify."    → Check it worked
GROWTH 🌱  "Learn."     → Save lessons, patch skills

Applying to Common Tasks

Training a model

DREAM  → Train sakthai-plus-1.5b on H100
HOPE   → Script ready? Token valid? Repos exist? GPU available?
CARE   → Run hf jobs uv run with correct params
JOY    → Push adapter and merged model to HF Hub
TRUST  → Verify model card, check downloads, run eval
GROWTH → Save training time, update cost estimates

Fixing a bug

DREAM  → What error? When did it start? What changed?
HOPE   → Check logs, find root cause, plan fix
CARE   → Fix the code, test locally
JOY    → Commit and push
TRUST  → Verify fix in production
GROWTH → Add test case, update troubleshooting guide

Creating a dataset

DREAM  → What gap? What format? What tools?
HOPE   → Check existing data, design schema, plan generation
CARE   → Generate examples, validate format, fix edge cases
JOY    → Push to HF Hub, update dataset card
TRUST  → Verify load_dataset works, check row counts
GROWTH → Document generation method, save script