Skip to content
Agentmart
Submit a tool

humanlayer-rpi-workflow

Trigger: RPI, research design implement, design discussion.

Skill by ramiro

IndexedApache-2.0
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/ram4-dev-humanlayer-rpi-workflow

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Trigger: RPI, research design implement, design discussion. Run a finite task through research, design approval, outline, implementation, and verification.

Use cases: design, research, productivity

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/humanlayer-rpi-workflow

What's inside

Skill name
humanlayer-rpi-workflow
Description
Trigger: RPI, research design implement, design discussion. Run a finite task through research, design approval, outline, implementation, and verification.
License
Apache-2.0
Metadata
author: ramiro · version: 1.1
Instructions
50 lines in SKILL.md · view on GitHub

Files (3)

  • SKILL.md
  • independent-outline-review.md · assets/
  • rpi-task-packet.md · assets/

SKILL.md (first lines)

## Activation Contract

Load this skill when a finite task needs current-state research and one integrated design discussion before implementation.

## Hard Rules

- Inspect the repository before asking questions.
- Keep research questions about current state; do not smuggle solution choices into research.
- Ground research in code, tests, commands, and primary sources.
- Ask one consequential design question at a time by default. Offer a recommendation and tradeoff.
- Before every approval request, follow `../_shared/approval-requests.md`: give a short bullet list explaining each approval item and each unresolved question.
- Do not create the structure outline until controlling design questions are resolved and the current design artifact is approved.
- After all controlling questions are answered and the complete outline is drafted, run an independent Pi review with `nan/glm5.3-flash` and reasoning `high` before requesting outline approval.
- Persist Pi's review in `05-independent-review.md`. Critical or controlling findings require an outline revision and another independent review; Pi never grants human approval.
- Outline feedback updates the outline only. It never authorizes implementation.
- Implement in vertical slices with unit, integration, static, build, E2E, and manual evidence as applicable.
- Before implementation or any code mutation, load `worktree-first-development` and create a fresh Worktrunk worktree. The implementation gate is not active until the worktree path and branch are recorded.

## Decision Gates

| Phase | Gate |
|---|---|
| Research questions | Scope covers the unknown current state |
| Research | Evidence answers the questions or names exact gaps |
| Design discussion | Explicit decisions and approval bound to current revision |
| Structure outline | Each slice has an observable result and checks; Pi review is pending |
| Independent review | Pi reviewed the exact outline revision; controlling findings are resolved |
| Implementation | Previous slice verified; manual checks remain pending until confirmed |

## Execution Steps

1. Create the artifacts in `assets/rpi-task-packet.md` under `.agent-workflow/tasks/<slug>/`.
2. Complete research questions, then research.
3. Write current state, desired state, options, tradeoffs, and open questions in the design discussion.
4. Stop for approval. Supersede approval if the design revision or scope changes.
5. Write the complete structure outline after all controlling questions are answered.
6. Use `herdr-development-orchestration` to run an independent Pi review on `nan/glm5.3-flash`, reasoning `high`; record scope, exact reviewed revision, findings, dispositions, session reference, and cleanup in `05-independent-review.md`.
7. Revise and re-review until no critical or controlling finding remains, then stop for explicit human outline approval.
8. Create the approved slice's fresh Worktrunk worktree, then implement and verify it. Record the worktree identity and any deviations in the controlling artifact.

## Output Contract

Return the phase, active gate, artifact changed, evidence gathered, Pi review state and exact reviewed revision, approval state, checks run, manual checks pending, and next authorized action.

## References

- `assets/rpi-task-packet.md`
- `assets/independent-outline-review.md`

From ram4-dev/shared-skills, Apache-2.0.

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 23, 2026 (6 days ago)

From GitHub (fetched Oct 24, 2026).

Details

Version
1.1
License
Apache-2.0
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from ramiro

Cite this page

humanlayer-rpi-workflow skill. Agent Mart. https://agentmart.dev/skills/ram4-dev-humanlayer-rpi-workflow (last updated 2026-10-24)