Green Claude plugin
Eco-design rules and a local audit for Claude Code: less resource use in code, database queries and token spend.
- 87 ★ on the repository
- Last commit
Minimal access
Why this level
- Optional hooks run the audit script after the agent writes code
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/green-claudeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- Optional hooks run the audit script after the agent writes code
This is third-party code. Review the repository files before installing.
What it does
Green Claude steers Claude Code toward code that uses fewer resources. The ruleset comes from the Institut du Numerique Responsable: 107 general rules, 127 rules for 24 languages and frameworks, 13 responsible-use practices for AI and 30 token sobriety rules. The rules draw on RGESN 2024, GR491, the Green Software Foundation and the W3C Web Sustainability Guidelines. The `eco-audit.sh` script checks a file with local patterns and no model call, and reports findings with a category, an RGESN reference and advice, for example about heavy libraries, SQL queries and whole-library imports. `/green-claude` shows the full checklist. Optional hooks run the audit after the agent writes code.
Who it is for. Developers and teams who are required or wish to reduce service load and watch resources against RGESN.
Good fit when
- A new service is designed with eco-design in mind, and the agent needs ready rules.
- An existing file must be checked for heavy libraries, needless queries and dead code.
- Token use in a project should drop by following a list of practices.
- The check should arrive automatically, without a separate request to the agent.
Not a fit when
- An RGESN compliance certificate is needed: the audit flags candidates for review and certifies nothing.
- Real energy savings must be measured: the tool does not compute them.
- Rule application must be guaranteed: the skill guides the model, but compliance in every response is not assured.
Example request
Run an eco-design audit of api.js, show findings with advice and fix only the most significant ones.Limitations
The audit runs through `bash` and needs `jq`. Hooks are installed only by the `install.sh` script, and the Claude Code plugin manager does not configure them. For 38 rules the reference points to a specific RGESN criterion, for 67 only to a family, and the rest follow the Green Software Foundation. For Claude.ai and the API the skill is built into an archive named `green-coding` by `package-skill.sh`, because a name containing claude is rejected there. Audit results are interpreted with Claude, which spends tokens.
How to disable. Remove the plugin with `/plugin uninstall green-claude@green-claude` or delete `~/.claude/skills/green-claude`, and take the hooks out of Claude Code settings.
FAQ
How do I install it?
Through the plugin manager: `/plugin marketplace add Institut-du-Numerique-Responsable/green-claude` and `/plugin install green-claude@green-claude`. The hook variant comes from `install.sh` in a clone.
How do I audit a file?
Ask Claude for an eco-design audit of the file or run `bash skills/green-claude/scripts/eco-audit.sh api.js`. The script prints findings with a level, category and advice.
Does the audit call a model?
No, detection uses local patterns. Tokens are spent only when you go through results with Claude.
How do I use it in Claude.ai and through the API?
Build the archives with `bash skills/green-claude/scripts/package-skill.sh`, which writes into `dist/`. In Claude.ai the archive is uploaded in the skills section, and the API has its own archive.
Which standards does it follow?
RGESN 2024, GR491, Green Software Foundation principles and the W3C Web Sustainability Guidelines.
Discussion
No questions or reviews yet.
Related
PluginsA Claude Code plugin that extracts behavioral specs, test vectors and acceptance criteria from a codebase, with a provenance trail
A Claude Code plugin that generates and maintains CLAUDE.md: a 0 to 100 quality score, a 150-line cap and stale reference checks
A Claude Code plugin: 39 commands for PRs, review, commits and specs, nine expert roles and hooks, in eight language versions
Claude Code plugins: plan, work and review commands, subagents, hooks and writing skills in a practical minimalism spirit.
An installer giving Claude Code rules for phases, decision logs, debugging and tests in Go, Rust, Python and six more languages.
A Claude Code plugin that hands a task to a local no_human and reviews a branch with a reviewer that never saw the coder's session
README in short
Covers installation through the plugin manager, `install.sh` and archives for Claude.ai and the API, ways to use it, an audit example and rules by RGESN family.