/issue-mr · Workflow

Issue & MR

From a fuzzy idea to an issue + branch + MR ready to code.

FR EN

"/issue-mr I have uncommitted changes that touch both form validation and the cache layer, they need to be separated"

The skill detects the two sets of files, proposes splitting them into two separate issues/branches/MRs, and validates the split with you before creating anything. It discovers project conventions live (labels, default branch, commit scopes) and produces artifacts that never mention AI.

A.

What it does

Three behaviors: SCAFFOLD (clean tree to issue + branch + MR/PR shell), ISOLATE (uncommitted work gets isolated into its own issue/branch/MR, file list and message validated first), ANALYSE (a vague task to code exploration, design settled with you, a structured issue-spec — finding, why, scope, plan — ready to feed feature-loop).

B.

When to use it

When framing a task. Discovers project conventions live (labels, default branch, commit scopes) instead of hardcoding them, validates everything with you BEFORE creating anything, and never mentions AI. Not for implementing (feature-loop) nor for wrapping up an existing branch (branch-wrap-up).

The stance

The stance: never hardcode a project's conventions. The skill discovers labels, default branch and commit scopes live, and validates everything with you before creating anything. It never mentions AI in the artifacts it produces.

Install this skill

claude code
$ /plugin marketplace add ohugonnot/claude-skills
$ /plugin install issue-mr@web-developpeur-skills
#git#gitlab#github#issue#merge-request#workflow

Frequently asked questions

When do I use /issue-mr instead of /feature-loop?

/issue-mr structures and frames the task (issue, branch, MR shell). /feature-loop implements it. They pair naturally: /issue-mr first to create the issue-spec, /feature-loop next to execute it.

What does ANALYSE mode actually do?

It starts from a vague request, explores the code to understand the real scope, co-designs the approach with you, then produces a structured issue-spec ready for implementation.

Why doesn't the skill hard-code project conventions?

Conventions vary across projects (GitLab vs GitHub labels, main vs master branch, commit scopes). Discovering them live avoids silent mistakes and makes the skill reusable without any upfront configuration.

See the skill source: SKILL.md on GitHub. The skill also triggers automatically when your request matches its description, and an AI agent discovers it via the MCP server. To understand how these skills are designed, read the official skills patterns.

Feedback & discussion · 0

No feedback yet. Tried it? Tell us what you think.