mcpField
W
writing-plans

writing-plans

github.com/hjewkesAgent workflows443 installsUpdated 35w ago
Repository

About writing-plans

writing-plans is a skill for producing implementation plans as organized directories, with an orchestration plan and task-specific briefings for agents. It specifies plan structure, formats, lifecycle guidance, and handoff to execution.

Key features

  1. 01Plan directory structure
  2. 02Orchestration plan and task briefings
  3. 03Machine-readable task manifest
  4. 04Plan format and lifecycle guidance
  5. 05Execution handoff options

Use cases

  1. 01Preparing implementation work for agents
  2. 02Breaking work into task-specific briefings
  3. 03Handing a completed plan to execution

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/hjewkes/agent-skills --skill writing-plans

After installing, type/writing-plansin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

references/
briefing-template.md
formats.md
plan-lifecycle.md
scripts/
verify-criteria

Details

From the project README

Writing Plans

Overview

Write implementation plans as a plan directory — a tight orchestration plan plus per-task briefing files for agents. Assume agents have zero codebase context. Document everything they need: which files to touch, complete code, testing, validation criteria. DRY. YAGNI. TDD. Frequent commits.

Announce at start: "I'm using the writing-plans skill to create the implementation plan."

Context: Can be invoked standalone or as a handoff from brainstorming. Works in the current directory or a dedicated worktree.

Save plans to: .claude/plans/YYYY-MM-DD-<feature-name>/

See references/plan-lifecycle.md for directory lifecycle (creation, execution, cleanup, stale detection).

Plan Directory Structure

.claude/plans/YYYY-MM-DD-<feature-name>/
  plan.md              # Orchestration plan (<200 lines)
  manifest.json        # Machine-readable task/wave metadata
  briefings/
    task-01.md         # Per-task agent briefing
    task-02.md
    ...

Format Specifications

See references/formats.md for plan.md template, manifest.json schema, and briefing authoring rules.

Briefing file template: references/briefing-template.md.

Execution Handoff

After saving the plan directory, offer execution:

"Plan complete and saved to .claude/plans/<plan-id>/. How would you like to execute it?"

Options:

  1. Execute now — I'll dispatch agents per task with review between each (stays in this session)
  2. Execute in separate session — Open a new session and invoke plan-execution for batch execution

REQUIRED SUB-SKILL: Use plan-execution for either option.

Comments

Markdown supported