LYGO Automation Workflows v1.0.0

LYGO edition of the solopreneur automation playbook — rebuilt for the lattice:

• Narrow triggers (not every “save time” chat)
• Privacy / least-privilege / consent warnings first
• Local-first preference: Sandcastle · Continuum · mint · PDW · Ollama army · self-hosted n8n
• Advisor only — does not connect Zapier/Make/CRM or move data for you

Signature: Delta9Phi963-AUTOMATION-WORKFLOWS-v1.0.0
Inspired by community skill jk-0001/automation-workflows (credit); LYGO hardening is original.

When to use

• Designing LYGO steward / agent automations with audit trails
• Auditing repetitive lattice ops (ledger rebuilds, witness packs, heartbeat touches)
• Comparing Zapier / Make / n8n after local options, with privacy constraints
• Emitting a consent-aware workflow plan JSON via the local planner CLI

When NOT to use

• Vague “automate my life” without a concrete recurring task
• Secretly wiring payment/CRM webhooks without steward review
• Auto-publishing to social / ClawHub / HF / live Star Chart
• Any request that needs this skill to execute third-party APIs (it won’t)

Install

npx clawhub@latest install deepseekoracle/lygo-automation-workflows
python scripts/self_check.py
python scripts/workflow_planner.py demo

FULL unlocked engineer zip: SkillHub #full-lygo → lygo-automation-workflows-full.zip

Local planner CLI (optional)

# Score a task
python scripts/workflow_planner.py audit-task \
  --name "Rebuild PDW ledger" --minutes 10 --frequency-per-month 30 --repetitive

# Emit a plan (stdout); write needs --i-consent
python scripts/workflow_planner.py plan \
  --name "Form to local CRM" \
  --trigger "New form submission" \
  --action "P0-gate payload" \
  --action "Store minimal fields locally" \
  --field email --field name \
  --tool lygo-sandcastle --tool n8n-self-hosted

Step 1 — Identify what to automate (LYGO lens)

Automate tasks you do ≥2×/week that are repetitive + rule-based and don’t need creative judgment.

Time cost: (minutes × frequency_per_month) / 60 hours/month.

Good lattice candidates

• Scheduled heartbeat / deadman touch
• Digest/witness packs from known local files
• Continuum seal after a known folder of claims
• Mint → anchor snippet generation (human still posts)
• Rebuild ledgers / run self_checks

Keep manual

• Champion persona replies that need nuance
• Live Star Chart accepts
• Anything touching raw payment/PII until a privacy design exists

Step 2 — Choose tools (local-first)

Step 3 — Design (with privacy)

TRIGGER: …
CONDITIONS: … (+ P0 / validation)
ACTIONS: …
DATA FIELDS: list every field moved (minimize)
VENDORS: list every system that receives data
ERROR HANDLING: alert steward — never silent fail
DISABLE PATH: how to turn it off + who can audit logs
CONSENT: who approved connecting each vendor

Hard rules

• No secrets in Slack/email alerts
• Prefer redacted notifications
• Document retention
• Human remains the publisher for social/lattice posts

Step 4 — Build & test

1. Dry-run / validate each step
2. Test empty fields + garbage input (P0)
3. Confirm failure alerts fire
4. Only then enable on a schedule

Step 5 — ROI (same math, LYGO values)

Payback months ≈ setup cost / monthly time-value saved. Prefer payback < 3 months, unless the automation unlocks safety (e.g. deadman heartbeat).

Mistakes to avoid (LYGO)

• Automating a broken process
• Over-broad skill triggers (“automate everything”)
• No error handling
• Moving PII/payments across vendors without consent review
• Skipping Continuum/mint receipts for consequential runs

Pair with

Security

See references/SECURITY.md and references/SKILLSPECTOR_AUDIT.md.

Δ9Φ963 — automate the repetitive · protect the sensitive · consent before vendors · human publishes.