Personal PM

Overview

Personal PM is a local-first workflow for planning with a coding agent. The reusable skill gives Codex, Claude Code, Gemini CLI, or another agent a process for reading goals, projects, and unfinished work, then drafting one daily priority and a few support tasks. A local web app makes the plan and completion history easier to review.

The loop is simple: goals become a daily plan, completion feedback is recorded, and that context informs the next run. Unfinished work stays visible so I can revisit its scope and next action.

Everything stays local-first. My real goals, tasks, and logs live under private/, while the reusable `SKILL.md`, app, validators, templates, and demo data sit outside it so they are safe to share.

Why I Built It

Built this for my own use to turn larger personal goals into concrete daily tasks.

I often want to make progress on side projects or learning goals, but those can get delayed or forgotten when the next step is not clear. This skill helps me convert overall and weekly goals into realistic daily tasks with one clear priority and a few actionable support tasks.

I find visual feedback useful, so the analytics view helps me review progress and spot repeated misses. The system is Markdown-first, so the planning files are easy to store, edit, and review locally.

What I’m Testing

• Can an agent turn broad goals into a small, actionable daily plan?
• Does carrying unfinished work forward with context help it suggest a clearer next step?
• Can completion history help calibrate how much work goes into the next plan?

These are the questions behind the workflow. The daily plans and completion records give me material to review; they are not a formal evaluation of agent performance.

How It Works

• Set goals, active projects, and weekly focus in simple Markdown files
• Use the guided setup to draft a weekly focus with Codex, Claude Code, or Gemini CLI
• Run the personal-pm skill to create one clear priority and a few realistic support tasks
• Carry unfinished work forward with context so it becomes a smaller next action instead of disappearing
• Review projects, completion trends, and repeated misses in the local app

What I Review

I look at the proposed priority, the size of each support task, and how carried-forward work is rewritten. After the day, completion records and the 90-day analytics view help me check for tasks that keep getting missed and plans that may be too ambitious.

The public demo uses sample planning data. My own goals, daily tasks, and feedback stay in the private workspace.

Workflow Components

• One clear daily priority: each plan starts with one main P1 and a few support tasks
• Plans that adapt: completion history and unfinished work shape how big the next plan is and when tasks land
• Guided weekly setup: an agent drafts a weekly focus from current goals, projects, and a few setup answers
• Progress visibility: project views and 90-day analytics show completion patterns and repeated misses
• Private but shareable: real planning data stays private while the skill, app, templates, validators, and demo data remain public-safe

Screenshots

Enlarge image ↗

Enlarge image ↗