with one click
automation-best-practices
Sub-skill of automation: Best Practices.
Menu
Sub-skill of automation: Best Practices.
| name | automation-best-practices |
| description | Sub-skill of automation: Best Practices. |
| version | 1.0.0 |
| category | development |
| type | reference |
| scripts_exempt | true |
# Store workflow definitions in git
git add automation/workflows/
git commit -m "feat: Add data sync workflow"
# Use environment-specific configurations
production:
schedule: "0 */6 * * *"
parallelism: 10
development:
schedule: null # Manual trigger only
parallelism: 2
# Add observability to all workflows
def with_monitoring(workflow_name):
start_time = time.time()
try:
result = execute_workflow()
metrics.record_success(workflow_name, time.time() - start_time)
return result
except Exception as e:
metrics.record_failure(workflow_name, str(e))
alerts.send(f"Workflow {workflow_name} failed: {e}")
raise
# Document workflow purpose and dependencies
# ABOUTME: Syncs customer data from CRM to data warehouse
# ABOUTME: Depends on: crm-api, warehouse-connection
# ABOUTME: Schedule: Every 6 hours
Karpathy's LLM Wiki — build and maintain a persistent, interlinked markdown knowledge base. Ingest sources, query compiled knowledge, and lint for consistency.
Operate ace-linux-1 as the continuous AI-agent control surface for overnight and continuous batches that keep GTM material moving toward client outreach.
Canonical GitHub issue planning route — issue intake, strengthened resource intelligence, repo-tracked plan artifact, adversarial review, GitHub progress posting, future-issue capture, explicit approval gate before execution, machine-dispatch readiness checks, and execution-ready delegation packaging for Claude agent teams.
Build a docs-only execution handoff bundle after a completed implementation wave — follow-up issue drafts, scoped authorization note, deploy checklist, operator note, copy/paste command bundle, and incremental commit hygiene.
Audit and dispose of session residue (orphan files, scratch dirs, sibling-repo state, locks, trash-stages) BEFORE claiming a task complete. Required gate before any agent says "all done", "task complete", or hands work back to user/orchestrator.
Plan and operate a Hermes-led control plane that routes AI provider work across workstations using quota urgency, machine readiness, GitHub issue gates, and a dispatch ledger.