Agent Skills: Task Prioritization Skill

Task prioritization and scoring

UncategorizedID: aaaaqwq/claude-code-skills/task-prioritization

Install this agent skill to your local

pnpm dlx add-skill https://github.com/aAAaqwq/AGI-Super-Team/tree/HEAD/skills/task-prioritization

Skill Files

Browse the full folder contents for task-prioritization.

Download Skill

Loading file tree…

skills/task-prioritization/SKILL.md

Skill Metadata

Name
task-prioritization
Description
Task prioritization and scoring

Task Prioritization Skill

Manually run the Task Prioritization agent to recalculate priority scores for all active tasks.

When to use

  • User asks to "update task priorities"
  • User asks "what are my top priorities" (run this first, then show results)
  • After bulk task changes (many deadlines changed, new blockers added, etc.)
  • To preview priority changes before they auto-run

How to execute

Step 1: Run agent

cd $AGENTS_PATH/task-prioritization

# Preview changes (recommended first)
python3 task_prioritization_agent.py --dry-run --verbose

# Apply changes
python3 task_prioritization_agent.py

Step 2: Review output

Agent will show:

  • Total tasks processed
  • Number of significant changes
  • Top 5 tasks by priority
  • Telegram notification if critical changes

Step 3: Show user results

Read top tasks from CSV and present:

# Read updated priorities
head -20 $PM_PATH/pm_tasks_master.csv

Sort by priority_score and show top 5-10 tasks.

Expected behavior

  • Updates priority_score and last_updated columns in pm_tasks_master.csv
  • Creates backup before write
  • Sends Telegram alert if hot tasks in top 3
  • Logs execution to pm_priority_agent_log.csv

Error handling

If agent fails:

  • Check /tmp/task-prioritization-agent-error.log
  • Verify pm_tasks_master.csv exists and is readable
  • Check if backup file exists: pm_tasks_master.csv.bak

Related skills

  • show-today — show today's tasks (uses priority_score)
  • daily-briefing — morning report (uses priority_score)

Notes

  • Agent runs automatically 3x/day (08:00, 13:00, 18:00 Mon-Fri)
  • Manual runs are safe (idempotent)
  • Use --dry-run to preview without changes
  • Algorithm details in agent README.md