npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

cgemini

v0.3.1

Published

Conductor + Gemini: use Gemini as an independent code reviewer inside any Conductor workspace — CLI, MCP tool, and /gemini slash command.

Readme

cgemini

cgemini = Conductor + Gemini.

A CLI and MCP server that puts Gemini inside any Conductor.build workspace as an independent code reviewer. Claude writes the code; Gemini reviews it — different model, no shared context, different blind spots caught.

npm install -g cgemini
cgemini login
cgemini review --staged          # Gemini reviews your staged changes
cgemini review --last --focus security
git diff main | cgemini review

Why

Conductor runs many coding agents in parallel, each in its own workspace. cgemini gives every one of those agents access to Gemini — configured once in ~/.cgemini/config.json, available everywhere — as a second opinion that didn't write the code it's reviewing.

Install

npm install -g cgemini

Requires Node.js 18 or newer (uses the built-in fetch).

Setup

Get an API key from Google AI Studio, then:

cgemini login                              # interactive, hidden input
# or
cgemini config set-key <your-key>          # non-interactive

Optional: pick a default model.

cgemini config set-model models/gemini-2.5-pro

Check what's active:

cgemini config show

Usage

cgemini "Summarize git diff in one line"
git diff | cgemini "What changed?"
cgemini "Write a haiku about Friday"

Only the model's response text is written to stdout. Errors and status messages go to stderr. That makes it safe to pipe, capture, or use in Conductor's shell-command blocks.

Code review

Ask Gemini to independently review your changes — useful because Gemini didn't write the code and has no context bias.

cgemini review --staged                   # review what's staged for commit
cgemini review --last                     # review the last commit
cgemini review --last --focus security    # security-only pass
cgemini review --last --focus logic       # logic/correctness only
git diff main | cgemini review            # review a branch diff
cgemini review src/auth.js                # review specific file (vs HEAD)

Focus options: all (default) · security · logic · style

The review ends with a clear one-line verdict: LGTM / LGTM with minor notes / Needs changes.

Commands

| Command | Description | | --- | --- | | cgemini "<prompt>" | Send a prompt, print the response | | cgemini review | Review a git diff with Gemini (independent code reviewer) | | cgemini login | Interactively save your API key | | cgemini config set-key <key> | Save an API key non-interactively | | cgemini config set-model <model> | Save the default model | | cgemini config show | Print effective config (key is masked) | | cgemini config remove-key | Remove the stored API key | | cgemini install | Register the MCP server and /gemini slash command with Claude Code | | cgemini --version | Print the installed version | | cgemini --help | Print usage |

Environment overrides

All config values can be overridden per-invocation without touching the saved config:

| Variable | Purpose | | --- | --- | | CGEMINI_API_KEY | API key (takes priority over the stored key) | | GEMINI_API_KEY | Alternate API key variable, also honored | | CGEMINI_MODEL | Model name (e.g. models/gemini-2.5-pro) | | CGEMINI_CONFIG_DIR | Override the config directory (default ~/.cgemini) |

Resolution order: env var → config file → built-in default.

Models

The default is models/gemini-2.5-flash. If the selected model returns a transient error (429/5xx) or is unavailable, cgemini retries briefly and then falls back through this list:

models/gemini-2.5-flash
models/gemini-2.5-flash-lite
models/gemini-2.0-flash-001
models/gemini-2.0-flash-lite-001
models/gemini-2.5-pro
models/gemini-pro-latest

Authentication errors (401/403) fail immediately rather than retrying.

Using cgemini from Conductor

cgemini plugs into any Conductor.build workspace without per-project setup. One install registers two surfaces with Claude Code at the user scope (~/.claude/...), so every Conductor agent inherits both your Gemini key and the integration.

cgemini login       # one-time: store your key in ~/.cgemini/config.json
cgemini install     # one-time: register MCP server + /gemini slash command

That's it. Open any Conductor workspace and you have two ways to use Gemini:

Gemini as independent code reviewer (primary use case)

The most valuable workflow in Conductor: Claude writes the code, Gemini reviews it. Different model, no shared context, catches different blind spots.

Via MCP tool — ask the agent:

"Use the review_code tool to review your changes before committing."

The agent runs git diff --cached, passes it to Gemini, reads the review, and fixes any issues before they ship.

Via slash command — type directly in the Conductor prompt window:

/gemini review --staged

Or set it up as part of your agent's task instructions:

"After making changes, run git diff --cached | cgemini review and fix any issues Gemini flags before proceeding."

1. ask_gemini MCP tool (agent-driven)

cgemini install runs claude mcp add gemini -s user -- cgemini-mcp, which gives the Claude Code agent inside every Conductor workspace a new tool. Ask the agent naturally:

"Get a second opinion from Gemini on this refactor." "Use ask_gemini to summarize the changes in this diff."

The agent decides when to call it and weaves the response into its reasoning — same pattern as Context7 or Linear.

2. /gemini slash command (user-driven)

cgemini install also writes ~/.claude/commands/gemini.md. Type this in any Conductor prompt window:

/gemini Explain blockchain in simple terms

Claude Code shells out to cgemini and injects the response inline. Good for quick direct questions without engaging the agent's reasoning loop.

Can cgemini appear in Conductor's model-selection dropdown?

No. Conductor runs Codex + Claude Code agents; its model selector reflects what those two runtimes support (Anthropic + OpenAI models). There's no extension API to register a third agent in the dropdown. The MCP-tool-plus-slash-command approach above gives Gemini the same reach inside Conductor without needing that slot.

Install flags

| Flag | Effect | | --- | --- | | cgemini install | Install both surfaces (default) | | cgemini install --mcp-only | Register MCP server only | | cgemini install --slash-only | Write slash command only (no claude CLI needed) | | cgemini install --force | Overwrite an existing ~/.claude/commands/gemini.md |

Development

git clone https://github.com/wakqasahmed/cgemini
cd cgemini
node bin/cgemini.js --help
npm test

Publishing

This repo auto-publishes to npm when you push a semver tag:

npm version patch          # or minor / major
git push && git push --tags

See .github/workflows/publish.yml. You'll need an NPM_TOKEN secret on the GitHub repo.

License

MIT — see LICENSE.