flow-next-ralph-init_skill

This skill scaffolds or updates a repo-local Ralph harness to enable opt-in autonomous coding workflows.

Python

458

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill gmickel/gmickel-claude-marketplace --skill flow-next-ralph-init

SKILL.md4.0 KB

Overview

This skill scaffolds or updates a repo-local Ralph autonomous harness under scripts/ralph/. It safely copies templated scripts and helper tools into the repository and preserves existing configuration when updating. Use it to opt in to Ralph mode for overnight autonomous coding with built-in review backend selection. The workflow is designed to be local, minimal, and reversible.

How this skill works

The skill detects the repository root, checks for an existing scripts/ralph/ directory, and sets update or fresh-install mode accordingly. For fresh installs it copies all template files and helper executables into scripts/ralph/, makes key scripts executable, and creates a runs/ folder. For updates it backs up config.env, replaces template files while preserving runs/ and restores the config file. During fresh installs it also probes and prompts to choose an available review backend.

When to use it

You want a repo-local autonomous Ralph harness for overnight or batch coding.
You need to install or refresh Ralph scripts after plugin updates.
You want to preserve existing Ralph settings while updating templates.
You want a local workflow that selects a review backend automatically or interactively.
You need a reproducible local entrypoint for Ralph runs (one-shot or loop).

Best practices

Run the skill from within the git repository root so repo root resolution is correct.
If updating, keep a copy of scripts/ralph/config.env outside the repo as extra backup.
Choose the review backend that matches your tools (RepoPrompt for visual macOS flows, Codex CLI for cross-platform GPT review).
Test with ./scripts/ralph/ralph_once.sh before running the full loop with ./scripts/ralph/ralph.sh.
Re-run the skill after plugin updates to refresh local scripts and helpers.

Example use cases

Initialize Ralph in a Python repo to run overnight coding passes with plan-first workflows.
Update an existing Ralph harness to pick up template improvements while keeping custom config.env.
Install a reproducible local flow entrypoint for CI or developer machines using the flowctl helper.
Switch a fresh repo to use a specific review backend after confirming available CLIs.

FAQ

No. During updates the skill backs up config.env, replaces templates, then restores your config.env so your custom settings are preserved.

How do I choose a review backend?

On fresh installs the skill detects available tools and prompts you to choose RepoPrompt, Codex, or none. If both are present it asks you to pick; otherwise it auto-selects the available option.