task-query

name	task-query
description	Query, list, and report on tasks in the s9 database
license	MIT
compatibility	opencode
metadata	{"audience":"all-agents","workflow":"task-discovery"}

What I Do

I provide comprehensive instructions for finding, listing, and reporting on tasks in the s9 task database using the task_show and task_list tools. Use this skill to discover available work, view task details, and generate reports.

Tool Overview

This skill uses:

• task_show tool - Get full details for a specific task
• task_list tool - List and filter tasks by various criteria
• task_report tool - Generate formatted reports

All tools return clean JSON results and automatically receive mission context from OpenCode.

List Available Tasks

By Status

task_list(status="TODO")
task_list(status="UNDERWAY")
task_list(status="TODO,UNDERWAY")
task_list(status="COMPLETE")

Available statuses:

• TODO - Not started
• UNDERWAY - In progress
• BLOCKED - Can't proceed
• PAUSED - Temporarily stopped
• REVIEW - Awaiting review
• COMPLETE - Finished
• ABORTED - Cancelled

By Role

task_list(role="Engineer", status="TODO")
task_list(role="Administrator", status="TODO")
task_list(role="Tester")

Available roles:

• Administrator
• Architect
• Engineer
• Tester
• Documentarian
• Designer
• Inspector
• Operator
• Historian

By Priority

task_list(priority="CRITICAL,HIGH", status="TODO")
task_list(priority="HIGH")
task_list(priority="MEDIUM,LOW")

Priority levels:

• CRITICAL - Immediate action required
• HIGH - Important, do soon
• MEDIUM - Nice to have
• LOW - Do when time permits

Active Tasks Only

Show only TODO and UNDERWAY tasks:

task_list(active_only=True)

This is the most common query for finding work.

By Agent

See tasks claimed by specific agent:

task_list(agent="Goibniu")
task_list(agent="Ishtar", status="UNDERWAY")
task_list(agent="Ptah", status="COMPLETE")

Combining Filters

Filters can be combined:

# High priority Engineer tasks that are TODO
task_list(role="Engineer", priority="HIGH", status="TODO")

# Active tasks assigned to Goibniu
task_list(agent="Goibniu", active_only=True)

# Completed tasks for a specific role
task_list(role="Tester", status="COMPLETE")

View Task Details

Get full details for a specific task:

task_show(task_id="ENG-H-0037")

Shows:

• Task ID, title, objective
• Full metadata (status, priority, role, category)
• Agent assignment (if claimed)
• Timestamps (created, claimed, closed)
• Time tracking (actual hours)
• Objective and description
• Dependencies
• Progress notes
• File path to markdown artifact

Example Output

{
  "task_id": "ENG-H-0037",
  "title": "Implement Rate Limiting Middleware",
  "status": "UNDERWAY",
  "priority": "HIGH",
  "role": "Engineer",
  "category": "Security",
  "agent": "Goibniu",
  "created_at": "2026-02-03T10:00:00+00:00",
  "claimed_at": "2026-02-05T14:00:00+00:00",
  "actual_hours": 4.0,
  "objective": "Add rate limiting to protect API endpoints from abuse",
  "description": "Implement token bucket rate limiting with configurable limits per endpoint",
  "dependencies": [],
  "notes": [
    "Implemented token bucket algorithm, added configuration (2.0 hours)",
    "Added tests, all passing (4.0 hours)"
  ],
  "file_path": ".opencode/work/tasks/ENG-H-0037.md"
}

Generate Reports

Markdown Report

task_report(format="markdown")

Generates markdown-formatted report of all tasks, suitable for documentation.

Table Format

task_report(format="table")

Generates ASCII table suitable for terminal display.

JSON Format

task_report(format="json")

Generates JSON output for programmatic processing.

Filter Reports

Reports can be filtered by role:

task_report(role="Engineer", format="markdown")
task_report(role="Administrator", format="table")

Finding Work

Find Available Work for Your Role

# Find TODO tasks for your role
task_list(role="Engineer", status="TODO")

# Find high-priority TODO tasks
task_list(role="Engineer", priority="CRITICAL,HIGH", status="TODO")

# Find any active work
task_list(role="Engineer", active_only=True)

Find Critical/Urgent Work

# Critical tasks across all roles
task_list(priority="CRITICAL", status="TODO")

# High priority tasks needing attention
task_list(priority="HIGH", status="TODO,UNDERWAY")

Find Blocked Tasks

# See what's blocked
task_list(status="BLOCKED")

# Check your blocked tasks
task_list(agent="YourName", status="BLOCKED")

Check Dependencies

Before claiming a task, check if it has dependencies:

result = task_show(task_id="ENG-H-0038")
# Check result["dependencies"]

# Check if dependency is complete
dep_result = task_show(task_id="ENG-H-0037")
# Check dep_result["status"] == "COMPLETE"

Common Query Patterns

Morning Standup

# What am I working on?
task_list(agent="YourName", status="UNDERWAY")

# What's next to work on?
task_list(role="YourRole", priority="HIGH", status="TODO")

# What's blocked?
task_list(status="BLOCKED")

Finding Next Task

# 1. Check critical tasks first
task_list(priority="CRITICAL", status="TODO")

# 2. Then high priority for your role
task_list(role="YourRole", priority="HIGH", status="TODO")

# 3. Look at medium priority if nothing urgent
task_list(role="YourRole", priority="MEDIUM", status="TODO")

When to Auto-Claim vs. Ask

Auto-claim after finding when the user gives an imperative command:

• "Find the next unclaimed task" → Find AND claim automatically
• "Get the next task" → Find AND claim automatically
• "Claim the next task" → Find AND claim automatically
• "What should I work on next?" → Find AND claim automatically

Ask before claiming when the user is inquiring:

• "What tasks are available?" → Show list, then ask if they want to claim
• "Show me unclaimed tasks" → Show list, then ask if they want to claim
• "List tasks for [role]" → Show list, then ask if they want to claim

The key distinction: imperative commands ("find", "get", "claim") indicate the user wants you to take action. Inquiry commands ("what", "show", "list") indicate the user wants information first.

Progress Check

# How many tasks completed?
task_list(status="COMPLETE")

# How many tasks in progress?
task_list(status="UNDERWAY")

# How much work left?
task_list(status="TODO")

Team Visibility

# What is each agent working on?
task_list(status="UNDERWAY")

# What has been completed?
task_list(status="COMPLETE")

# Generate status report
task_report(format="markdown")

Tips and Best Practices

Do

• ✅ Use active_only=True for quick work discovery
• ✅ Check dependencies before claiming tasks
• ✅ Filter by role and priority to focus
• ✅ Use task_show to understand task fully before claiming
• ✅ Generate reports for team status updates
• ✅ Auto-claim when user gives imperative commands ("find next task", "get next task")
• ✅ Distinguish between inquiry ("what tasks?") and action requests ("find next task")

Don't

• ❌ Don't list all tasks without filters (too much noise)
• ❌ Don't ignore CRITICAL tasks
• ❌ Don't claim tasks without checking dependencies
• ❌ Don't forget to check who's working on what
• ❌ Don't ask "Would you like me to claim this?" after imperative commands
• ❌ Don't auto-claim when user is just browsing tasks

Output Format

List Output

Task lists return JSON arrays with key information:

[
  {
    "task_id": "ENG-H-0037",
    "title": "Implement Rate Limiting",
    "status": "UNDERWAY",
    "priority": "HIGH",
    "role": "Engineer",
    "agent": "Goibniu"
  },
  {
    "task_id": "ENG-H-0038",
    "title": "Configure Gateway",
    "status": "TODO",
    "priority": "HIGH",
    "role": "Operator",
    "agent": null
  }
]

Show Output

Full task details including:

• All metadata
• Full objective and description
• Complete notes history
• Time tracking
• Timestamps
• File path

Troubleshooting

"No tasks found"

• Check your filters aren't too restrictive
• Try task_list() without filters to see all tasks
• Verify tasks exist for that role/status/priority

"Task not found"

• Check task ID spelling and case
• Use task_list() to find available tasks
• Make sure you're using the full task ID (e.g., ENG-H-0037)

"Invalid filter value"

• Check spelling of status/priority/role
• See valid values sections above
• Use exact capitalization

Performance Tips

• Use filters to reduce output
• Use active_only=True instead of listing all statuses
• Use task_show for single task details
• Generate reports periodically, not constantly

See Also

Related Skills:

• task-create - Creating new tasks
• task-claim - Claiming tasks found through queries
• task-update - Updating tasks after claiming
• task-close - Closing tasks when complete
• task-management - Overview of task system

Documentation:

• .opencode/data/README.md - Complete s9 system reference