Skip to content
Back to skills

Plan Writer

ASecurity

Decomposes a feature spec into a task list. Writes plan.md with each task tagged frontend|backend|mixed (plus optional owner/content) and given a [TODO] status marker the orchestrator updates as work progresses.

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 19, 2026
developmentfrontendbackenddocumentation

Security analysis

A100/100

Pro scans all 4 files and shows the line behind each finding

Scanned October 19, 2026

npx -y skills add petruseugene/full-development-cycle --skill plan-writer --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Plan Writer?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Plan Writer
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/petruseugene-plan-writer/badge)](https://www.skillsdirectory.com/skills/petruseugene-plan-writer)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: plan-writer
disable-model-invocation: true
description: Decomposes a feature spec into a task list. Writes plan.md with each task tagged frontend|backend|mixed (plus optional owner/content) and given a [TODO] status marker the orchestrator updates as work progresses.
---

# Plan Writer

## When to use

- Feature flow step 5, after spec-writer (and wireframer if applicable).
- Skipped when `feature.json.separate_plan == false`: spec-writer then appends the tasks to `spec.md` using the same `references/task-format.md`.

## Required reading

1. Read `${CLAUDE_PLUGIN_ROOT}/skills/playbook/SKILL.md` — the playbook core. Read it before anything else.
2. `references/task-decomposition.md`
3. `references/sizing-heuristics.md`
4. `references/task-format.md`
5. `<feature folder>/spec.md`
6. `<feature folder>/wireframes/index.html` (if exists)
7. `<feature folder>/feature.json` (for `design_source`, `has_ui`, `has_backend`)

## The job

1. Read the spec's requirements and acceptance criteria.
2. Produce ordered tasks. Each task:
   - Has a clear, observable end state (something a reviewer can check)
   - Is sized 30-90 minutes of implementer time (see `references/sizing-heuristics.md`)
   - Is tagged with one stack token `frontend`, `backend` or `mixed`, optionally plus `owner` and/or `content`, comma-separated (e.g. `(backend, content)`, `(owner)`). `owner` = manual owner step, never dispatched, still needs `Touches` and `Depends on`. `content` needs `**Recipe:** <name>` from `profile.recipes`. Both: track `full` only (see `references/task-format.md`)
   - References specific files to create/modify (paths, where known)
   - Has step-level breakdown only when an implementer would benefit (small tasks can be one paragraph)
   - **For every AC the task covers, if that AC references a `wireframes/*.html` file, the task MUST include a `**Wireframes:**` line listing every such file as a relative path from the feature folder.** This is what the orchestrator forwards into the implementer's dispatch prompt — omit it and the visual contract leaks. If a task covers no wireframe-anchored ACs, omit the field entirely.
   - **MUST include a `**Touches:**` line** — a comma-separated glob list (best-effort estimate) of files the implementer is expected to write to. The orchestrator uses this to compute parallel groups via disjoint-globs partitioning. Prefer narrow globs over broad ones. List shared files explicitly (e.g., `package.json`, `tsconfig.json`, shared type modules) when they will be touched. See `references/task-decomposition.md` for rules.
   - **MUST include a `**Depends on:**` line** — a list of prerequisite task IDs whose output this task reads (types, schemas, components, endpoints), or the literal word `none`. Asymmetric — defines tier ordering, not file overlap.
   - **MAY include a `**Size:**` line** (`S|M|L`, after `**Depends on:**`). Missing means `M`. Size is informational only and never changes the model. See `references/task-decomposition.md`.
   - **Merge small tasks.** Merge Size-S tasks in the same tier that share a file owner (overlapping `Touches`, or the same single directory) into one task. Never merge owner tasks, and never merge tasks with a different stack tag. See `references/sizing-heuristics.md`.
3. Write `plan.md` with this frontmatter and heading, followed by the `## Tasks`, `## Execution plan` and `## Task details` sections exactly as defined in `references/task-format.md` (including the note that the Execution plan overview is documentation only):
   ```markdown
   ---
   feature: <slug>
   spec: ./spec.md
   status: [TODO]
   ---

   # <Feature> — Plan
   ```

## Output contract

```yaml
---
status: DONE | DONE_WITH_CONCERNS | NEEDS_CONTEXT | BLOCKED
output_file: docs/features/<folder>/plan.md
summary: <N tasks; first task: <task 1 name>>
tokens_used: ~<N>k
duration: <Xm Ys>
---
```

Final message: the YAML block above plus at most 5 lines of prose (playbook core §6).

Files in this skill

  • SKILL.md3.9 KB
  • references/sizing-heuristics.md1.2 KB
  • references/task-decomposition.md4.3 KB
  • references/task-format.md2.5 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.