ワンクリックで
task-runner
Pick and start the next task from @tasks. Queries status, presents available work, sets up context, and transitions to DOING.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Pick and start the next task from @tasks. Queries status, presents available work, sets up context, and transitions to DOING.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Project orientation, status overview, and skill/agent discovery. Shows metrics, blockers, available skills, agents, and quick-start actions.
Methodology for complex changes spanning 3+ files, vendored code, or cross-crate boundaries. Consult before any multi-file change.
Creating plans with 5+ items, phased checkpoints, orchestration
When and how to use visual regression tests and AI-readable screenshots.
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
| auto-generated | true |
| name | task-runner |
| description | Pick and start the next task from @tasks. Queries status, presents available work, sets up context, and transitions to DOING. |
The Task Runner skill helps you pick the next task to work on. It queries the @tasks PRD system, presents available work grouped by category and priority, helps you choose, sets up context, and transitions the task to DOING.
Invoke with /task.
Run the emacsclient dashboard and blocked-items queries to understand the current state:
# Get overall dashboard metrics
emacsclient -s sakya -e '(prd-dashboard-cli)'
# Get blocked items specifically
emacsclient -s sakya -e '(prd-list-blocked)'
If emacsclient is unavailable (server not running, not installed), fall back to direct file reads:
@tasks/{projects,bugfixes,improvements}/*.org files directlyITEM, DOING, REVIEW, DONE, BLOCKED) from headings:PROPERTIES: drawers for DEPENDS, AGENT, EFFORT, PRIORITYRead task files from all three categories:
@tasks/projects/PROJ-*.org - Feature projects@tasks/bugfixes/BUG-*.org - Bug fixes@tasks/improvements/IMP-*.org - Improvements and refactorsFor each file, identify ITEM entries (pending, unblocked tasks). An ITEM is unblocked if:
DEPENDS property, ORDEPENDS have status DONEGroup available tasks and present them clearly:
For each available task, show:
| Field | Source |
|---|---|
| ID | CUSTOM_ID property (e.g. ITEM-042) |
| Title | Heading text |
| Category | Parent file (PROJ/BUG/IMP) |
| Priority | PRIORITY property |
| Agent | AGENT property (linked agent name) |
| Effort | EFFORT property |
| Dependencies | DEPENDS property (show status of each) |
| Files | FILES property (what will be touched) |
Start the presentation with a summary: "N tasks available: X projects, Y bugfixes, Z improvements"
Offer selection guidance based on different strategies:
Ask the user which task they want to work on. Accept task ID (e.g. "ITEM-042") or description keywords.
Once a task is selected, prepare the working context:
@tasks/agents/<agent>.orgFILES or COMPONENT_REF properties exist, read those source files to understand current statePresent a brief summary of the context gathered.
After the user confirms they want to start:
ITEM to DOING in the org fileemacsclient -s sakya -e '(prd-validate-file-cli "@tasks/path/to/file.org")'
If validation fails, show the errors and fix them before proceeding.
If the user invokes /task with an argument (e.g. /task ITEM-042), skip steps 3-4 and jump directly to setting up context for that specific task.