ai4pkm-helper_skill

Overview

This skill helps configure and manage the AI4PKM orchestrator, workers, and CLI workflows for Obsidian vault automation. It guides you through creating agents, setting up multi-worker comparisons, and integrating pollers and cron schedules. The goal is reliable, testable automation for content ingestion, enrichment, and task management.

How this skill works

The skill inspects and edits orchestrator.yaml to add or update nodes, defaults, pollers, and worker definitions. It guides CLI usage for status, logs, and agent control, and recommends test steps to validate triggers and outputs. Multi-worker setups are supported to run the same task across different executors and store separate outputs for comparison.

When to use it

• Add, modify, or remove agents/nodes in orchestrator.yaml
• Configure multi-worker execution to compare AI models
• Update orchestrator defaults like executor, timeout, or max_parallel
• Add or tune pollers for external data sync (limitless, apple_photos, apple_notes)
• Run or troubleshoot CLI commands for orchestrator management

Best practices

• Validate YAML syntax and keep a backup before saving orchestrator.yaml
• Test new agents with a small input set before enabling full automation
• Give each worker a distinct output_path to avoid result mixing
• Use clear labels for workers and consistent prompt filenames in _Settings_/Prompts
• Monitor logs (ai4pkm --logs) and check status (ai4pkm --status) after changes

Example use cases

• Create a file-triggered agent that watches Ingest/Clippings and writes enriched notes to AI/Articles
• Set up multi-worker comparison to run the same summarization prompt with Claude and Gemini and save outputs separately
• Enable limitless poller to sync lifelog entries into Ingest/Limitless on a periodic interval
• Add a cron-based daily driver workflow to generate a journal roundup every hour at :15 and :45
• Use CLI to list agents, run a single agent for testing, and inspect logs when execution fails