How do I install the Local Gemma Worker skill?

Run npx skillmds@latest add newmindsgroup/local-gemma-worker in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Local Gemma Worker safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Local Gemma Worker?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Local Gemma Worker free to use?

Yes. Installing skills from SkillMD is free. This skill is licensed under MIT.

Who published Local Gemma Worker?

newmindsgroup (@newmindsgroup) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Local Gemma Worker by newmindsgroup

Name: Local Gemma Worker
Author: newmindsgroup

Local Gemma Worker

Delegate bounded local/offline tasks to Google's Gemma 4 through Ollama from Codex. Use when the user asks to use Gemma, assign work to Gemma, run a local model pass, do private/offline summarization, extraction, first-pass review, or compare drafts before Codex synthesizes the final answer.

File contents

Local Gemma Worker

Use this skill when the user asks Codex to assign a bounded task to local Gemma or to use a private/offline local model as a helper.

Command

Use codex-gemma, which calls the local Ollama API at http://127.0.0.1:11434.

Profiles:

codex-gemma --fast "task": gemma4:e4b-it-q8_0 and the current default installed worker
codex-gemma --balanced "task": gemma4:26b-a4b-it-q8_0
codex-gemma --deep "task": gemma4:31b-it-q8_0

Input patterns:

Pipe text, files, or diffs into the command.
Use --image /absolute/path.png for image inspection.
Use --json only when a downstream script needs JSON-shaped output.
Run codex-gemma --check before relying on it if model availability is uncertain.
Run codex-gemma-install-full to resume/pull all three configured profiles when network time is available.

Optional Codex CLI profiles are also available for explicit full-agent local runs:

codex exec --profile gemma-local-fast
codex exec --profile gemma-local-balanced
codex exec --profile gemma-local-deep

The Codex profiles use local Ollama aliases such as codex-gemma-fast to avoid noisy model-name telemetry warnings in Codex CLI runs.

Good Delegation Targets

Summarization
Extraction
Classification
First-pass code review
First-pass strategy or document critique
Brainstorming variants
Draft comparison
Private/offline document analysis

Boundaries

Gemma is a helper, not the Codex runtime. Do not delegate final judgment, file edits, command execution, live web research, or verification of facts that Codex can directly verify.

The helper refuses -cloud model tags by default so local/private requests remain local. Only override with CODEX_GEMMA_ALLOW_CLOUD=1 when the user explicitly asks for a cloud model.

After Gemma responds, Codex must critically read the result, verify important claims when practical, and synthesize the final answer itself. Do not imply Gemma changed files or validated behavior unless Codex separately confirmed it.

---name: local-gemma-workerdescription: Delegate bounded local/offline tasks to Google's Gemma 4 through Ollama from Codex. Use when the user asks to use Gemma, assign work to Gemma, run a local model pass, do private/offline summarization, extraction, first-pass review, or compare drafts before Codex synthesizes the final answer.license: MIT---# Local Gemma WorkerUse this skill when the user asks Codex to assign a bounded task to local Gemma or to use a private/offline local model as a helper.## CommandUse `codex-gemma`, which calls the local Ollama API at `http://127.0.0.1:11434`.Profiles:- `codex-gemma --fast "task"`: `gemma4:e4b-it-q8_0` and the current default installed worker- `codex-gemma --balanced "task"`: `gemma4:26b-a4b-it-q8_0`- `codex-gemma --deep "task"`: `gemma4:31b-it-q8_0`Input patterns:- Pipe text, files, or diffs into the command.- Use `--image /absolute/path.png` for image inspection.- Use `--json` only when a downstream script needs JSON-shaped output.- Run `codex-gemma --check` before relying on it if model availability is uncertain.- Run `codex-gemma-install-full` to resume/pull all three configured profiles when network time is available.Optional Codex CLI profiles are also available for explicit full-agent local runs:- `codex exec --profile gemma-local-fast`- `codex exec --profile gemma-local-balanced`- `codex exec --profile gemma-local-deep`The Codex profiles use local Ollama aliases such as `codex-gemma-fast` to avoid noisy model-name telemetry warnings in Codex CLI runs.## Good Delegation Targets- Summarization- Extraction- Classification- First-pass code review- First-pass strategy or document critique- Brainstorming variants- Draft comparison- Private/offline document analysis## BoundariesGemma is a helper, not the Codex runtime. Do not delegate final judgment, file edits, command execution, live web research, or verification of facts that Codex can directly verify.The helper refuses `-cloud` model tags by default so local/private requests remain local. Only override with `CODEX_GEMMA_ALLOW_CLOUD=1` when the user explicitly asks for a cloud model.After Gemma responds, Codex must critically read the result, verify important claims when practical, and synthesize the final answer itself. Do not imply Gemma changed files or validated behavior unless Codex separately confirmed it.

Frequently asked questions

Local Gemma Worker

File contents

Local Gemma Worker

Command

Good Delegation Targets

Boundaries

Related skills

Frequently asked questions

How do I install the Local Gemma Worker skill?

What does the Local Gemma Worker skill do?

Is Local Gemma Worker safe to use?

Which AI agents work with Local Gemma Worker?

Is Local Gemma Worker free to use?

Who published Local Gemma Worker?