Skip to content

spenserhale/human-driven-development

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Human-Driven Development (hdd)

A de-hooked clone of the Agentic Development Skills workflow, booted explicitly and kept human-in-the-loop at every gate.

spec → plan → implement → verify, with the human approving the design, reviewing the spec, choosing the execution path, and picking the final merge/PR outcome.

Unlike the original, there is no SessionStart hook — nothing is auto-injected. You start the workflow on demand:

/human-driven-development add react to this project

The human-driven-development skill is a short router (progressive disclosure): it holds the workflow map and loads each sub-skill on demand via the Skill tool.

The workflow

human-driven-development   (router — invoked explicitly via /human-driven-development)
  │
  ├─ hdd-brainstorming ─────────►  spec doc   docs/hdd/specs/YYYY-MM-DD-<topic>-design.md
  │     └─ hdd-writing-plans ───►  plan doc   docs/hdd/plans/YYYY-MM-DD-<feature>.md
  │           └─ execute the plan, task-by-task:
  │                • hdd-subagent-driven-development   (fresh subagent + 2-stage review per task)
  │                • hdd-executing-plans               (inline, with checkpoints)
  │                     ├─ hdd-test-driven-development        (red → green → refactor, per task)
  │                     ├─ hdd-requesting-code-review         (review per task / before merge)
  │                     ├─ hdd-verification-before-completion (evidence before any "done")
  │                     └─ hdd-finishing-a-development-branch (merge / PR / keep / discard — you choose)
  │
  ├─ hdd-using-git-worktrees      (opt-in — only if the harness/setup/project recommends it)
  └─ hdd-systematic-debugging     (any bug / test failure — root cause before fixes)

Human gates (kept from the original, unchanged)

  • Design approval — brainstorming presents the design section-by-section and waits for approval.
  • Spec review — you review the written spec before planning starts.
  • Plan review + execution choice — you pick subagent-driven vs. inline execution.
  • Merge/PR choice — finishing presents the 4-option menu (merge / PR / keep / discard) and waits.

For the fully-unattended counterpart that decides everything itself and opens a PR, use agentic-driven-development (/agentic-driven-development).

Install (npx skills)

npx skills add ./HumanDrivenDevelopment

Skills install flat, so every skill is prefixed hdd- to avoid colliding with the add- skills from Agentic-Driven Development when both are installed.

Conventions

  • Skills live in skills/<name>/SKILL.md (YAML frontmatter + markdown body).
  • Skills reference each other by their prefixed name, hdd-<skill>.
  • Workflow artifacts go in docs/hdd/specs/ and docs/hdd/plans/ (user preferences override).

Ported from obra/superpowers via Agentic Development Skills.

About

Human-in-the-loop agentic dev workflow: /human-driven-development brainstorms, seeks approval at every gate, and lets you choose merge/PR. Skills for Claude Code & other agents.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors