base/.agents/skills/task-management/SKILL.md

name, description

name	description
task-management	Manage, query, create, triage, research, and update project tasks using scripts/ and docs/tasks.md. Use when reading, creating, editing, prioritizing, or updating tasks, checklists, or project progress for any kind of project.

Task Management

Read the script guide and the tracker. Use its configured prefix, project keys, statuses, and types, not an assumed project or software stack. Examples below use task numbers so they work with any prefix.

Environment

The optional CLI tools require Python 3.9+ and Bash or Zsh:

source ./scripts/env.sh

Keep local overrides in ignored scripts/env.sh.local or .env. These are trusted shell scripts. TASKS_FILE chooses the tracker; TASK_ID_PREFIX overrides its prefix. All task commands accept -f /path/to/tasks.md.

Querying

./scripts/get-tasks
./scripts/get-tasks --all
./scripts/get-tasks --planning
./scripts/get-tasks --accept "Status:Ready"
./scripts/get-task 1
./scripts/get-task 1 --json
./scripts/get-task 1 --description-only

Creating and Updating

Read valid project/type/status values from frontmatter before invoking:

./scripts/add-task "Task Title" --project example-project --type task --status triage -d "Description and review criteria"
./scripts/update-task 1 --status in_progress
./scripts/update-task 1 --title "Updated Title"
./scripts/update-task 1 --append-description "- [ ] Additional criterion"
./scripts/update-task 1 --check-all
./scripts/update-task 1 --status testing

Use --epic or --subproject only with values defined in frontmatter. Numbers and configured prefixed IDs are accepted.

Synchronization and Review

• Anchor attributes in Detailed Tasks are the canonical task metadata.
• Do not manually edit generated summary/enum tables. Run ./scripts/update-tasks to regenerate them.
• Run ./scripts/get-task <number> after edits to verify persisted values.
• planning / triage / research scope work; ready permits execution; in_progress tracks active work; testing awaits human verification.
• testing applies to any project, including writing and research. It need not involve automated tests. Humans confirm done.
• Optional triage / research helpers require a supported external agent CLI; normal task management does not.
• If the CLI runtime is unavailable, explain that limitation. With user agreement, maintain canonical task details in Markdown and ensure tables remain consistent; do not silently claim script verification succeeded.