Embla Claude Plugins

A Claude Code plugin marketplace for the Embla organisation. It ships shared skills that other Embla repos install to get AI-assisted developer workflows: Jira integration, PR review, branch/commit conventions, sprint planning, deployment, Claude Code setup audits, and GDPR compliance.

Contents

• Plugins
• Installation
• Getting started
• Plugin reference
• Repository layout
• Contributing
• License

Plugins

Plugin	Version	What it gives you
embla-core	1.6.1	End-to-end dev lifecycle: Jira, story development, PR review, deploy, sprint planning, CI pipelines
project-setup	0.3.1	Audit a project's Claude Code setup, review session history, set up OpenWiki
grc-gdpr	1.0.4	GDPR compliance: code/system audits, privacy notices, DPAs, DPIAs, article-cited Q&A (vendored, MIT)

Install only the plugins you need. Each one adds its skill descriptions to every session's context.

Installation

1. Add the marketplace

From inside Claude Code in the consuming repo:

/plugin marketplace add EmblaTech/claude-plugins

2. Install plugins

/plugin install embla-core@embla-claude-plugins
/plugin install project-setup@embla-claude-plugins
/plugin install grc-gdpr@embla-claude-plugins

3. Share the setup with your team (recommended)

Commit the marketplace and enabled plugins to the consuming repo's .claude/settings.json so every developer gets them automatically:

{
  "extraKnownMarketplaces": {
    "embla-claude-plugins": {
      "source": { "source": "github", "repo": "EmblaTech/claude-plugins" }
    }
  },
  "enabledPlugins": {
    "embla-core@embla-claude-plugins": true,
    "project-setup@embla-claude-plugins": true
  }
}

Plugins auto-update when their version changes in this repo.

Getting started

1. Configure the project. Run /embla-core:config. It creates .claude/embla.json in your repo with the repo, Jira, sprint, branch-format, and team settings that the other embla-core skills read.
2. Check your Claude Code setup. Ask Claude to "audit my Claude Code setup". This runs project-setup:audit.
3. Start a story. Run /embla-core:develop EM-1234.

Plugin reference

embla-core

Skill	Invocation	Purpose
config	/embla-core:config	Create or update .claude/embla.json for a project
develop	/embla-core:develop <JIRA-ID>	Full story workflow: fetch, branch, brainstorm, TDD, commit, PR, then Done
jira	/jira spec <path>, /jira story, /jira bug …	Create Jira epics, stories, tasks, and bugs from a spec or free text
sprint-plan	/embla-core:sprint-plan	Plan a sprint in 4 phases: goal, stories, assignment, then creation in Jira
pr-review	/embla-core:pr-review <PR_ID> [--mode dev|lead|pipeline]	PR review with up to 7 parallel agents (code quality, security, performance, risk, coverage, dependency, requirement); dev/lead mode on Bitbucket, pipeline mode on any CI host
resolve-pr-feedback	/embla-core:resolve-pr-feedback <PR_ID>	Work through review comments after a rejection: explain, fix, commit, resolve, reassign
deploy	/embla-core:deploy <PR_ID>	Trigger CI/CD, poll the build, move Jira to Ready for QA
pipeline	/embla-core:pipeline or /embla-core:pipeline check	Set up or audit the CI pipeline file and .mcp.json for AI PR review on any host (Bitbucket and GitHub profiles, generated for others), plus Bitbucket-only ai-develop
init-llm-wiki	/embla-core:init-llm-wiki	Bootstrap a Karpathy-style LLM Wiki as a dev knowledge base
develop-auto	CI only	Headless version of develop, triggered by the ai-develop pipeline when a Jira comment tags @Claudedev

Configuration. Every embla-core skill reads .claude/embla.json in the consuming project. Each field resolves in this order: embla.json, then settings.json, then an env var, then a hardcoded default. The full schema is in plugins/embla-core/skills/config/SKILL.md.

Conventions enforced:

• Branches: feature/{JIRA-ID}/short-description, patch/{JIRA-ID}/short-description, release/{version} (formats are configurable per project in embla.json)
• Commits: Conventional Commits, e.g. feat(scope/EM-1234): short description

project-setup

Skill	Invocation	Purpose
audit	Ask to "audit my Claude Code setup"	Severity-tagged report covering settings placement, MCP health, permissions, plugin drift, stack coverage, and CLAUDE.md placement
session-review	/project-setup:session-review	Review session history for context, delegation, verification, and memory discipline, with quoted citations
setup-openwiki	/project-setup:setup-openwiki	Install LangChain OpenWiki's host-driven integration (no API key)

A config skill is planned. It will move here from embla-core.

grc-gdpr

Skill	Invocation	Purpose
gdpr-compliance	Mention GDPR, data protection, DPIA, etc.	Audit code and systems, draft privacy notices and DPAs, review data flows, answer questions with article citations

This plugin is vendored from an MIT-licensed upstream with local citation patches. Don't edit it in place. Follow the re-sync process in docs/grc-upstream.md.

Repository layout

.claude-plugin/marketplace.json      Marketplace manifest (lists all plugins)
LICENSE                              Apache-2.0 license for the repo
NOTICE                               Attribution for third-party code
plugins/<plugin-name>/
  .claude-plugin/plugin.json         Plugin name, version, description
  agents/*.md                        Named subagents (e.g. pr-review's 7 reviewers)
  skills/<skill-name>/
    SKILL.md                         Skill entry point (frontmatter: name, description)
    references/                      Supporting files, loaded only when SKILL.md says so
docs/                                Skill-authoring best practices, design specs/plans, upstream vendoring notes
.claude/                             This repo's own Claude Code settings and CLAUDE.md

Contributing

Contributor guidance lives in .claude/CLAUDE.md. The rules you're most likely to hit:

• Always bump the plugin version. Any change to a plugin must bump version in its plugin.json. Without the bump, consumers keep the old copy.
• New skill in an existing plugin: add skills/<name>/SKILL.md and bump the version. marketplace.json doesn't change.
• New plugin: add plugins/<name>/.claude-plugin/plugin.json, at least one working skill, and a marketplace.json entry, all in the same PR. Placeholder stubs aren't accepted.
• CI pins: generated pipelines clone this repo at a pinned commit. If a change to pr-review or develop-auto should reach existing pipelines, bump the pin in that skill's pipeline template: every hosts/*/template.yml for pr-review, workflows/ai-develop/template.yml for develop-auto. See "Updating the pinned plugin commit" in the pipeline skill's references/hosts/bitbucket/bitbucket-setup-guide.md.
• Use the right authoring skill: /writing-for-agents (from mattpocock-skills) for new and edited skills, /plugin-dev:create-plugin for new plugins.
• PR descriptions follow the format in CLAUDE.md: Summary, Motivation, Test Plan, Linked Issue.

License

Licensed under the Apache License, Version 2.0. Copyright 2026 Embla.

Two parts come from third parties, as credited in NOTICE:

• plugins/project-setup/skills/session-review/scripts/sessions.mjs is adapted from Anthropic's session-report plugin (Apache-2.0).
• plugins/grc-gdpr/ is vendored from an upstream project under the MIT License, which still applies to those files. See plugins/grc-gdpr/LICENSE.