
Hone
- 25 repo stars
- Updated June 30, 2026
- oskarhane/hone-ai
AI Coding Agent Orchestrator - PRD to tasks to implementation loop
About
hone is a Claude Code skill in the AI & Agent Building category. AI Coding Agent Orchestrator - PRD to tasks to implementation loop
- hone
- AI & Agent Building
- AI-coding skill
Hone by the numbers
- Data as of Jul 9, 2026 (Skillselion catalog sync)
/plugin marketplace add oskarhane/hone-ai/plugin install hone@hone-aiAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| repo stars | ★ 25 |
|---|---|
| Last updated | June 30, 2026 |
| Repository | oskarhane/hone-ai ↗ |
What it does
AI Coding Agent Orchestrator - PRD to tasks to implementation loop
README.md
hone - Claude Code Plugin
AI Coding Agent Orchestrator as a native Claude Code plugin. Run the full hone workflow (PRD → tasks → run → review → fix) directly inside Claude Code without spawning external processes.
Install
Add the marketplace and install the plugin:
/plugin marketplace add oskarhane/hone-ai
/plugin install hone@hone-ai
To update to the latest version:
/plugin marketplace update hone-ai
To test locally during development:
claude --plugin-dir ./claude-plugin
Skills
All skills are invoked via /hone:<skill-name>.
Workflow
| Skill | Description | Example |
|---|---|---|
/hone:auto |
Run the full chain (prd→tasks→run→review→fix) hands-free | /hone:auto "Add user authentication" |
/hone:agents-md |
Generate AGENTS.md project docs | /hone:agents-md --overwrite |
/hone:prd |
Generate PRD from feature description | /hone:prd "Add user authentication" |
/hone:prd-to-tasks |
Generate task YAML from PRD | /hone:prd-to-tasks .plans/prd-user-auth.md |
/hone:extend-prd |
Add requirements to existing PRD | /hone:extend-prd .plans/prd-user-auth.md "Add OAuth" |
/hone:run |
Execute implement/review/finalize loop | /hone:run .plans/tasks-user-auth.yml -i 5 |
/hone:review |
Strict end-of-feature audit of the branch | /hone:review |
/hone:fix |
Turn supplied issues (or "the above" from a prior review) into tasks and run them | /hone:fix .plans/tasks-user-auth.yml the above blocking issues |
/hone:pr |
Push the branch, open a PR, monitor CI checks, and auto-fix failures | /hone:pr .plans/tasks-user-auth.yml |
Info
| Skill | Description |
|---|---|
/hone:status |
Show incomplete task lists with progress |
/hone:prds |
List all PRDs with status |
/hone:prune |
Archive completed PRDs (--dry-run to preview) |
Common Workflow
Hands-free — one command runs prd→tasks→run→review→fix, asking batched questions once up front:
/hone:auto "Add user login with email and password"
Or step through each phase manually:
/hone:agents-md
/hone:prd "Add user login with email and password"
# Review .plans/prd-user-login.md
/hone:prd-to-tasks .plans/prd-user-login.md
/hone:run .plans/tasks-user-login.yml -i 10
/hone:review
/hone:fix .plans/tasks-user-login.yml the above blocking issues
/hone:prune
How It Works
Unlike the hone CLI (which spawns claude -p as subprocesses), this plugin runs everything natively inside Claude Code:
- PRD generation: Claude asks clarifying questions directly in conversation
- Task generation: Claude analyzes the PRD and produces task YAML
- Run loop: Each iteration launches a forked Agent that implements a task, gets it reviewed, and commits — all without subprocess overhead
- AGENTS.md: The source of truth for feedback commands (test, build, lint, format) — read at the start of every iteration
File Structure
.plans/
├── prd-<feature>.md # PRD documents
├── tasks-<feature>.yml # Task breakdowns
├── progress-<feature>.txt # Development logs
└── archive/ # Completed features
AGENTS.md # Project docs and feedback commands