From deck
Turn ideas into structured specs and execute them — conduct interviews to refine requirements, import existing plans, or run task lists via subagents. Use whenever starting a feature, planning implementation, or managing task execution.
How this skill is triggered — by the user, by Claude, or both
Slash command
/deck:specThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
```
/spec "idea" # Start interview from idea description
/spec @file.md # Start interview from file content
/spec run @SPEC-*.md # Execute tasks in specified SPEC file
/spec import @plan.md # Import Claude Code plan file to SPEC format
Route to the appropriate mode based on $ARGUMENTS:
Trigger: $ARGUMENTS starts with run
Example: run @SPEC-user-auth.md
Required: Must specify a SPEC file. If not provided, prompt user to specify one.
Trigger: $ARGUMENTS starts with import
Example: import @plan.md
Required: Must specify a plan file to import.
Trigger: All other cases
@file - Start from file content$ARGUMENTS directlyConduct adaptive interviews using AskUserQuestion. No fixed rounds — continue until the user signals completion.
Core Principles:
Topic Areas to Explore (adapt order and depth based on context):
Generate a short English slug: lowercase, hyphen-separated, 2-4 words (e.g., user-auth, dark-mode, api-rate-limit).
If plan mode is currently active, call ExitPlanMode before generating the SPEC file. The spec interview process itself serves as the planning phase — the SPEC document is the deliverable, not a plan file.
Ensure directory exists, then generate SPEC-{slug}.md using the template in references/spec-template.md. Adapt sections based on the interview — omit sections that are not relevant, expand those that are.
Import a Claude Code plan file and convert it to SPEC format for subagent execution. This enables isolated execution of each task with fresh context, avoiding context window overflow.
./plans/SPEC-{slug}.md and show summary| Plan Format | SPEC Format |
|---|---|
- [ ] Task | ⬜ Pending |
- [x] Task | ✅ Completed |
1. Task | ⬜ Pending |
| Nested items | Merged into Notes column |
Execute tasks sequentially using subagents. Main context handles scheduling and progress tracking only.
For each pending task in order:
last-updatedOn failure, stop and report: which task failed, error details, current progress, and how to retry (/spec run @SPEC-xxx.md resumes from the failed task).
npx claudepluginhub darknight/cc-deck --plugin deckCreates task specifications via codebase analysis, interactive clarifying questions on scope/constraints/edge cases, feature splitting checks, and agent delegation setup.
Orchestrates spec-driven development workflow (Requirements → Design → Tasks → Implementation) with approval gates. Activates for structured feature planning or 'use spec-driven'.
Conducts multi-round interviews to refine rough SPEC.md into complete, implementation-ready specifications with tasks. Use for new features, requirements refinement, or ideas to actionable specs.