Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Project Roadmap

ASecurity

Use when the user wants to start a new project, generate a project plan, create a roadmap, scaffold a PLAN.md and ROADMAP.md, or turn a project brief into a structured development plan with phases, tasks, exit criteria, and a live status tracker.

2 stars
0 votes
0 copies
0 views
Added 9/28/2026
ai-agentsgotestingapi

Works with

api

Security Analysis

A100/100

Scanned 9/28/2026

Install to Claude Code

$npx -y skills add TonyLikeDev/TrustMeBro --skill project-roadmap --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Project Roadmap?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Project Roadmap
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/tonylikedev-project-roadmap/badge)](https://www.skillsdirectory.com/skills/tonylikedev-project-roadmap)

More formats (shields.io, HTML) on the badges page.

Files
SKILL.md
---
name: project-roadmap
description: Use when the user wants to start a new project, generate a project plan, create a roadmap, scaffold a PLAN.md and ROADMAP.md, or turn a project brief into a structured development plan with phases, tasks, exit criteria, and a live status tracker.
---

# Project Roadmap Generator

You generate two documents for any project: a frozen master plan (`PLAN.md`) and a live status
tracker (`ROADMAP.md`). Together they solve two problems: wasted planning time at project start,
and context window pollution from stale history during development.

## Core Principle

- `PLAN.md` — the **master plan**: static, never edited to record progress. Defines objectives,
  phases, per-phase tasks (Build / Measure / Write), exit criteria, risks, and decisions.
- `ROADMAP.md` — the **live status board**: ticked as work happens. Progress bars, change log,
  immediate next actions. Never edit PLAN.md to record progress; that is ROADMAP.md's job.

## 3-Tier Context Loading Model

This skill produces artifacts consumed at three token-budget tiers:

| Tier | File | Max size | Loaded when |
|------|------|----------|-------------|
| 1 | `.bob/context/architecture.md` | ~500 tokens | Every session start — always in context |
| 2 | `.bob/context/current-phase.md` | ~200 tokens | Each session, replaces full ROADMAP.md |
| 3 | `PLAN.md` + `ROADMAP.md` | unbounded | Only when explicitly needed (planning/review) |

**Rule**: During execution (`/dev-workflow`), only Tier 1 + Tier 2 are loaded. Tier 3 is NEVER
loaded into the main context window — use `spawn_subagent` if a deep-read of PLAN.md is needed.

---

## Step 1 — Gather the Project Brief

Read the user's message for:
- **Project name and one-sentence goal**
- **Tech stack** (language, framework, platform)
- **Deliverables** (what "done" looks like — working software, report, demo, API, etc.)
- **Timeline** (days, weeks, or phases available)
- **Team size** (solo or multi-person; if multi, any machine/environment differences)

If any of these are missing and cannot be inferred, ask for them before proceeding. Do not
generate a plan for a vague brief — a narrow, specific plan is always more useful.

---

## Step 2 — Decompose into Phases

Break the project into 4–10 phases (or weeks if the user gave a weekly timeline). Each phase must:

1. Have a **single clear theme** (e.g. "Infrastructure", "Core Algorithm", "Testing", "UI", "Release")
2. List concrete **Build** tasks — specific files or modules to create
3. List **Measure** tasks — what to run, benchmark, or validate
4. List **Write** tasks — docs, reports, comments, changelogs
5. Have **Exit criteria** — a binary pass/fail statement that closes the phase

Use the following heuristics for phase sizing:
- A phase should be completable in 1–3 days of focused work (solo) or 1 week (team)
- If a phase has more than 8 Build tasks, split it
- The last phase is always "Polish, test, deliver" — never skip it

---

## Step 3 — Write PLAN.md

Create `PLAN.md` in the project root (or the path the user specified).

Structure:

```
# Project Plan: <name>

> Static master plan. Edit to refine the plan; never edit to record progress — use ROADMAP.md.

## Overview
- Goal: <one sentence>
- Stack: <language / framework / tools>
- Deliverables: <what done looks like>
- Timeline: <total duration>
- Team: <solo / members / machines>

## Decisions Log
| Topic | Option A | Option B | Decision | Reason |
(list any non-obvious architectural decisions made at planning time)

## Phase 1: <name>
### Build
- [ ] <specific file or module to create>
### Measure
- [ ] <what to run or validate>
### Write
- [ ] <doc or report to produce>
### Exit criteria
- [ ] <binary statement: tests pass / benchmark hits threshold / etc.>

## Phase 2: <name>
...

## Risks and Fallbacks
| Risk | Signal | Fallback |

## Research Questions (if applicable)
- RQ1: ...
```

Each task must be concrete enough that a developer can tick it without interpretation.

---

## Step 4 — Write ROADMAP.md

Create `ROADMAP.md` in the same directory as `PLAN.md`.

Structure:

```
# Roadmap and Status

Master plan: `PLAN.md`. This file is the live status board.
**Rule: tick items here in the same commit as the work. Never edit PLAN.md to record progress.**

Legend: `[x]` done · `[~]` written but not verified · `[ ]` open

Last updated: <today's date>

## Progress

<!-- progress:start -->
**Progress: 0% of <N>-phase plan** `[░░░░░░░░░░░░░░░░░░░░]`

| Phase | Items | Done | Progress | |
| :--- | ---: | ---: | :--- | ---: |
| Phase 1 | <N> | 0 | `[░░░░░░░░░░░░░░░░░░░░]` | not started |
...
<!-- progress:end -->

---

## Phase 1: <name>

### Build
- [ ] <copy from PLAN.md>

### Measure
- [ ] <copy from PLAN.md>

### Write
- [ ] <copy from PLAN.md>

### Exit criteria
- [ ] <copy from PLAN.md>

---

## Phase 2: <name>
(paste but leave all items [ ] — not started)

...

## Immediate Next Actions (in order)
1. <first concrete action to take right now>
2. <second action>
3. <third action>

## Change Log
- <today's date>: Plan generated by IBM Bob 2.0. Phase 1 not started.
```

---

## Step 5 — Create Tier 1: architecture.md

Create `.bob/context/architecture.md` (create `.bob/context/` if it does not exist).

This is the **always-loaded** Tier 1 anchor — loaded at the top of every session, before any
task execution. Keep it under **500 tokens**.

```
# Architecture Snapshot — <project name>

> Tier 1 context: always load this file at session start.
> Last updated: <date>

## What this project is
<2 sentences: goal + what it produces>

## Stack
- Language/runtime: <...>
- Key libraries/frameworks: <...>
- Storage/infra: <...>

## Key decisions (non-obvious ones only)
- <decision 1>: <why>
- <decision 2>: <why>

## Source layout (top-level only)
<src/
  module-a/   — <one-line purpose>
  module-b/   — <one-line purpose>>

## What NOT to re-read every session
- PLAN.md: full phase/task breakdown — use /roadmap-navigator instead
- ROADMAP.md: full status — use .bob/context/current-phase.md instead
```

---

## Step 6 — Create Tier 2: current-phase.md

Create `.bob/context/current-phase.md`.

This is the **session-scoped** Tier 2 snapshot. Keep it under **200 tokens**.

```
# Current Phase Context

**Project**: <name>
**Current phase**: Phase 1 — <name>
**Phase status**: not started
**Last updated**: <today's date>

## Open Tasks

### Build
- [ ] <Phase 1 build tasks>

### Measure
- [ ] <Phase 1 measure tasks>

### Write
- [ ] <Phase 1 write tasks>

### Exit criteria
- [ ] <exit criteria>

## Completed in this phase
0 of <N> tasks done

## Immediate next action
<first item from "Immediate Next Actions" in ROADMAP.md>

## Phase summary (for Bob's reasoning)
<2–3 sentences: what Phase 1 is about and why it comes first>
```

**Critical**: do NOT include completed phases, future phases, or the change log in this file.

---

## Step 7 — Summarise and Hand Off

After creating all files, tell the user:

1. What was created (file paths)
2. The phase breakdown (phase names and approximate size)
3. Token cost breakdown:
   - Tier 1 `architecture.md`: ~500 tokens (always loaded)
   - Tier 2 `current-phase.md`: ~200 tokens (session scope)
   - Full PLAN.md + ROADMAP.md: only loaded on demand via subagent
4. How to start working: *"Run `/roadmap-navigator` at the start of each session to refresh
   Tier 2. Run `/dev-workflow` to execute tasks."*
5. The one immediate next action from `ROADMAP.md`

Do not start executing the plan — that is the `dev-workflow` skill's job.

Attribution

TonyLikeDevTonyLikeDev
View sourceMore from TonyLikeDev →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

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 (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

1074701 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

695601 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

691 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →