Dev Progress Tracker

Maintain docs/todos.md as a living, project-level progress tracker that persists across sessions and AI platforms.

When to Use This Skill

• Session start: a docs/todos.md file exists — read it before doing any work
• Task completes: check off the finished item and update any metrics at the top
• New work identified: add it to the appropriate section rather than keeping it in the conversation
• New multi-phase project: docs/todos.md does not yet exist — create it following the structure below
• Milestone reached: update the dashboard scores/metrics to reflect current state

Trigger phrases: "check todos", "update todos", "where did we leave off", "what's next", "track progress", "progress dashboard", docs/todos.md

What This Skill Does

docs/todos.md is the single source of truth for ongoing project work. It is:

• Forward-looking only — completed sprints/phases belong in history files, not here
• Platform-agnostic — any AI agent on any platform reads and writes the same file
• Manually triggered — agents update it when explicitly asked or when the session-start rule fires; they do not rewrite it on every response

Structure

Every docs/todos.md follows this three-section layout:

Section 1 — Dashboard (always at top)

A metrics table that is updated after each sprint, push, or milestone. Customize columns for the project type.

# [Project Name] — Progress Dashboard

**Branch:** `<active branch>`

---

## Scores (update after each sprint)

| Metric | Current | Target | Delta |
|--------|---------|--------|-------|
| <metric 1> | <value> | <target> | <gap> |
| <metric 2> | <value> | <target> | <gap> |

Common metrics by project type:

• Software/testing: code coverage %, passing test count, CI score
• Feature work: tasks done / total, open bugs, blocked items
• Research/analysis: data processed, models evaluated, accuracy

Section 2 — Task Roadmap

Checkboxes organized by sprint, phase, or theme. Completed items are struck through or the section header is marked [DONE] and collapsed.

## Sprint / Phase N — [short description] (target: <metric delta>)

- [ ] Task description — specific, actionable, one sentence
- [ ] Task description
- [x] Completed task

## Sprint / Phase N+1 — [short description]

- [ ] Task description

Rules for tasks:

• Each task must be independently completable in one session
• Never nest tasks more than one level deep
• Write tasks in imperative mood ("Read X", "Write Y", "Fix Z")
• Include the file or module affected when relevant

Section 3 — Functionality Matrix (optional)

Use when the project has multiple feature dimensions that need coverage tracking. A table per dimension, with columns for tested/untested status.

## Functionality Matrix

### [Dimension Name]

| Feature | Tested? | File/Location | Sprint |
|---------|---------|--------------|--------|
| Feature A | ✅ Done | `path/to/file.py` | — |
| Feature B | ❌ Missing | — | Sprint N |
| Feature C | ✅ Partial | `path/to/file.py` | — |

Instructions

At Session Start

1. Run: does docs/todos.md exist?
   • Yes: read it now before doing anything else. Note the current sprint and any blocked tasks.
   • No: ask the user if they want one created before proceeding.
2. Briefly summarize where things stand (2-3 sentences max) so the user knows you have context.
3. Do not repeat the full file contents back to the user — just the relevant current state.

Checking Off Completed Tasks

1. Change - [ ] to - [x] for each task completed in this session.
2. If the task completion changes a dashboard metric (e.g., test count, coverage), update the dashboard table too.
3. If an entire sprint/phase is done, add [DONE] to its heading.

Adding New Tasks

1. Identify the correct sprint or phase section based on scope and priority.
2. Add the task as - [ ] <imperative description> with enough context to be actionable without re-reading the full conversation.
3. If a new sprint/phase section is needed, add it after the last existing one with a clear goal and target metric.

Creating the File from Scratch

When docs/todos.md does not exist:

1. Ask the user: "Should I create docs/todos.md to track this project? If so, what metric are we optimizing for (e.g., test coverage, features shipped, bug count)?"
2. Create docs/ if it does not exist.
3. Write the file with Section 1 (dashboard) and Section 2 (first sprint/phase tasks). Add Section 3 only if the project has multiple coverage dimensions.
4. Populate the dashboard with whatever current metrics the user provides — use ? for unknown values rather than leaving cells empty.
5. Confirm the file was created and show the dashboard table.

Keeping It Focused

• Do not bloat the file with meeting notes, decisions, or rationale — those belong in DEVLOG.md or ADRs.
• Do not move completed items to a "done" archive at the bottom — just mark them [x] and leave them in place. The file stays readable.
• Do not auto-update the file on every assistant turn — only update when something actually changed.

Format Rules

• File path: docs/todos.md (always this exact path)
• Heading level 1: project name + "Progress Dashboard"
• Heading level 2: major sections (Scores, Sprint N, Functionality Matrix)
• Tables: GitHub-flavored markdown pipe tables
• Checkboxes: - [ ] (pending), - [x] (done)
• Status emoji: ✅ Done, ❌ Missing, ✅ Partial (for functionality matrices only)
• No hard line wrapping inside table cells or bullet points

Related Skills

• session-history — captures what happened in a session (retrospective); todos.md is forward-looking
• code-commit-workflow — after committing, check off the relevant tasks in todos.md
• wrap-up-session — runs at end of session; reads todos.md as part of cleanup
• implementation-plan — produces the initial sprint breakdown that populates todos.md Section 2