Skip to main content
skill CX Codex

plan-task — Claude Code skill: turn a draft into a phased plan

Turns a rough task spec into a plan with acceptance criteria, an architecture overview and verifiable phases, each gated by a judge.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add NeoLabHQ/context-engineering-kit --skill plan-task -a claude-code
Open on GitHub

What plan-task does

Point it at a task file under .specs/tasks/draft/. The skill runs research, a pass over the codebase and a business-analysis stage at the same time; the last one tightens the description and writes acceptance criteria. Those results get merged into an architecture overview. Then the task is split into separate sub-task files, grouped into phases you can check on their own.

Behaviour is driven by flags. --target-quality sets the judge threshold, --max-iterations caps retries per phase, and --included-stages or --skip pick which stages run. --fast keeps only business analysis and decomposition while judges still score at a lower bar; --one-shot runs the same two stages with no judges at all. Use --refine to re-run only what a git diff of the task file affected, and --human-in-the-loop to pause after chosen phases.

The task file path is required and the file has to sit in .specs/tasks/draft/. When the run finishes, the task moves from draft/ to todo/. With --one-shot nothing filters out a weak plan, so that mode is the wrong pick for work you care about.

What it helps with

  • turn a rough task note into an implementable plan
  • break a large task into sub-task files by verifiable phase
  • get an architecture overview before writing code
  • rerun affected stages after the spec was edited
  • set a quality bar for automated phase checks

Example prompts for the agent

  • Refine .specs/tasks/draft/add-validation.feature.md with --target-quality 4.0
  • Rerun only the affected stages for .specs/tasks/draft/add-validation.feature.md using --refine
  • Pick up where it stopped for .specs/tasks/draft/add-validation.feature.md with --continue

How to install plan-task

1Skills CLI (Vercel)

npx skills add NeoLabHQ/context-engineering-kit --skill plan-task -a claude-code

2Manually: put the files in place

~/.claude/skills/plan-task/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/plan-task/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about plan-task

How do I install plan-task in Claude Code?

Run npx skills add NeoLabHQ/context-engineering-kit --skill plan-task -a claude-code. The skill then lives at ~/.claude/skills/plan-task/SKILL.md.

What does plan-task do?

It refines a draft spec through analysis, builds an architecture overview, splits the work into sub-task files grouped by phase, and promotes the task from draft/ to todo/.

Can I skip the quality checks?

Yes: --skip-judges or the --one-shot shortcut drops them. Phases then run without a judge, so a weak plan passes through unchecked.

Similar skills

All skills for Agent setup: memory, context, orchestration →

From the same set64

reflect A skill that makes the agent reread its own answer and list the flaws before you accept the work. CX Codex +1 1,400installs thought-based-reasoning A skill for Claude and other AI coders that gives the model a method for working through hard problems in explicit steps. Useful when a plain prompt returns the wrong answer. CX Codex +1 1,400installs prompt-engineering Guides prompt design for agents, subagents, and LLM integrations. Helpful when building production templates, chain-of-thought prompts, or agent hooks. CX Codex +1 1,400installs agent-evaluation Framework for testing agents, prompts, and skills using multi-dimensional scoring rubrics and LLM-as-judge prompts. CX Codex +1 1,300installs memorize Transfers insights from agent reflections and critique sessions into CLAUDE.md. Keeps your context file sharp by structuring lessons instead of bloating it with raw logs. CX Codex +1 1,300installs subagent-driven-development Runs an implementation plan through fresh subagents and puts code review between tasks, so context stays clean and issues surface early. CX Codex +1 1,300installs judge-with-debate The skill runs three independent judges who argue over a solution's score until they agree. Useful when a single pass feels too thin for a real verdict. CX Codex +1 1,300installs update-docs Updates project documentation based on uncommitted git changes or the latest commit. Keeps README files, docs/, and JSDoc comments synchronized with your codebase. CX Codex +1 1,300installs

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor