start-roadmap

A workflow executor for completing a roadmap, which is a plan containing multiple software tasks and their dependencies. It runs tasks in dependency order, meaning prerequisite work must finish before dependent work starts.

In plain words
What is it for?
Checking a roadmap's dependency tree, running planned tasks in order, and optionally running independent tasks in separate Git worktrees for isolation.
Why use it?
It prevents tasks from running before their required plans or dependencies are ready. It can continue through all runnable tasks without repeated confirmation.

Command for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/ayoubben18/ab-method/start-roadmap
Clone the repo
git clone --depth 1 https://github.com/ayoubben18/ab-method

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 645 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.00645
Opus 5 $0.00000 $0.00322
Sonnet 5 $0.00000 $0.00129
Haiku 4.5 $0.00000 $0.00064

Measured 2d ago against content hash c3ea4c81d6d7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

start-roadmap scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/commands/start-roadmap.md · 35 lines

How it starts

The opening of the file, as written. The whole thing — 35 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Start Roadmap

Description

Execute a roadmap autonomously to completion — the layer above /start-task. Verifies the dependency tree is planned, then runs each task in dependency order (each task like /start-task), with independent tasks optionally running in parallel. Executor only.

Usage

/start-roadmap [name]

Behavior

Loads and executes the start-roadmap workflow from .ab-method/core/start-roadmap.md.

It will:

  1. Always check .ab-method/structure/index.yaml first for paths.
  2. Read roadmap.md, build the DAG from depends-on edges.
  3. Dependency-tree verification gate (before anything runs): walk from the roots down; a task is runnable only if it has a plan and its deps are done. If a needed dependency task is unplanned, it stops loudly and tells you which task to /create-task first. It executes the fully-planned prefix and stops cleanly where planning runs out.
  4. Starts immediately — no "Proceed?" prompt. Parallelism comes from the invocation: sequential by default (dead safe), or git worktrees when you ask for it (true isolation + speedup for independent tasks). Then runs autonomously.
  5. Execute in dependency order; each task runs by /start-task rules (subagent per mission, tdd, commit per green mission). Runtime-adaptive shape: on Claude Code it nests by default (task-subagent → mission-subagents) for per-task context isolation; on Codex it stays flat (parent → mission-subagents), since Codex can't nest subagents. Unknown runtime falls back to flat. Update roadmap.md statuses as tasks complete.

Notes

  • Post-implementation review per task: each task runs by /start-task rules, so review-implementation runs on its diff after the last mission — safe fixes auto-applied, everything written to docs/tasks/<slug>/review.md. The critic subagents follow the same nest-on-Claude / flat-on-Codex rule as the mission subagents. The final report lists each task's review.md with its open-finding count.
  • Re-runs resume from roadmap.md — re-verifies and continues from the frontier.
  • Incremental / extend case: after a roadmap completes, use /extend-task to add missions to any task (it reopens that task in roadmap.md), then re-run /start-roadmap — it runs only the tasks that now have unchecked missions, in dependency order, and skips finished ones. A task "needs work" whenever its tracker has an unchecked mission, regardless of its status: flag. Common after long PRD-driven builds you want to tweak.
  • Stops loudly on red (never commits broken work), exactly like /start-task, one level up.

Read the full file on GitHub · 35 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 35 lines · 0 tokens per session scan A c3ea4c81d6d7

Subscribe to this mod's changes

start-roadmap is a command published in the GitHub repository ayoubben18/ab-method (187 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 645 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.