From rp1-dev
Orchestrates interactive loop for small, low-risk code changes via sub-agent delegation. Performs scope checks to redirect medium/large work to /build-fast or /build.
npx claudepluginhub rp1-run/rp1This skill is limited to using the following tools:
Interactive builder loop for rapid, small changes. Delegates each request to a single general sub-agent.
Orchestrates interactive implementation of small low-risk code changes via sub-agent delegation. Performs scope checks to redirect medium/large tasks to structured builders.
Offers lightweight pipeline for small routine changes (≤3 files, low-risk, e.g., typos, version bumps, lint fixes) via leader-builder-tester-ship, skipping full workflow.
Implements features from chat context or description via full agent workflow without tickets, boards, or status updates. Loads domain skills, plans, codes, tests, and opens PR.
Share bugs, ideas, or general feedback.
Interactive builder loop for rapid, small changes. Delegates each request to a single general sub-agent.
This command ONLY orchestrates. It does NOT implement code.
Extract these parameters from the user's input:
| Parameter | Required | Default | Description |
|---|---|---|---|
REQUEST | No | "" | Initial development request (may be empty; will prompt if missing) |
Environment values (resolve via shell):
RP1_ROOT: !rp1 agent-tools rp1-root-dir (extract data.root from JSON response)stateDiagram-v2
[*] --> GetRequest
GetRequest --> Clarify: vague
GetRequest --> ScopeCheck: clear
Clarify --> ScopeCheck: clarified
ScopeCheck --> Redirect: medium_or_large
ScopeCheck --> Build: small
Redirect --> Prompt
Build --> Prompt
Prompt --> Commit: user=commit
Prompt --> Build: user=refine
Prompt --> GetRequest: user=new
Prompt --> [*]: user=exit
Commit --> GetRequest
If REQUEST empty:
{% ask_user "What would you like to build?" %}
Super vague (ask for clarification):
Clear enough (proceed):
If vague: ask ONE clarifying question. Do NOT over-interrogate.
Before delegating, assess the request:
| Factor | Small (proceed) | Medium/Large (redirect) |
|---|---|---|
| Files | 1-3 | >3 |
| Systems | 1 | >1 |
| Risk | Low | Medium or High |
| Estimated effort | <2h | >2h |
If Medium or Large: Do NOT delegate. Instead output:
## This request is better suited for a structured workflow
**Request**: {summary}
**Why**: {brief reason — e.g. touches multiple systems, high risk, many files}
**Recommended**:
- For medium work (2-8h): `/rp1-dev:build-fast "{REQUEST}"`
- For large work (>8h): `/rp1-dev:build "{feature-id}"`
Then loop to §1.5 (Post-Build Prompt) so the user can submit a smaller request or exit.
Spawn a single general sub-agent to implement the request:
{% dispatch_agent "rp1-dev:express-builder" %} Implement the following change in the codebase:
{REQUEST}
Keep changes minimal and focused. Run any relevant lint/format/test checks after making changes. Do NOT commit. {% enddispatch_agent %}
Wait for completion. Do NOT implement anything yourself.
After builder completes:
{% ask_user "What would you like to do next?", options: "Commit & move on", "Refine", "New task (no commit)", "Exit" %}
| Option | Action |
|---|---|
| Commit & move on | Commit current changes (conventional commit), then loop to 1.1 |
| Refine | Ask what to change, re-invoke §1.4 with refinement as REQUEST |
| New task (no commit) | Loop to 1.1 without committing |
| Exit | STOP |
After a scope redirect (§1.3), show only "New task" and "Exit" options.
When user chooses "Commit & move on":
git add -A)Clear REQUEST, loop to 1.1 (Get Request).
On exit, report tasks completed count.
## Session Summary
**Tasks Completed**: {count}
Express session ended.
YOU MUST:
YOU MUST NOT: