https://github.com/rmedranollamas/research-cli
Gemini Deep Research Agent CLI
https://github.com/rmedranollamas/research-cli
deep-research gemini-cli gemini-cli-extension gemini-extension research-agent
Last synced: about 2 months ago
JSON representation
Gemini Deep Research Agent CLI
- Host: GitHub
- URL: https://github.com/rmedranollamas/research-cli
- Owner: rmedranollamas
- License: apache-2.0
- Created: 2026-01-01T22:52:43.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2026-02-16T18:35:05.000Z (about 2 months ago)
- Last Synced: 2026-02-16T18:57:46.076Z (about 2 months ago)
- Topics: deep-research, gemini-cli, gemini-cli-extension, gemini-extension, research-agent
- Language: Python
- Homepage:
- Size: 202 KB
- Stars: 12
- Watchers: 0
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Research CLI
Stateless CLI for Gemini Deep Research.
Refer to the [Instruction Manual](MANUAL.md) for detailed setup and usage guides.
## Installation
### As a Gemini CLI Extension (Recommended)
This is the fastest and most efficient way to install the tool. The Gemini CLI will automatically fetch the optimized, platform-specific binary from our latest release.
```bash
gemini extensions install rmedranollamas/research-cli
```
### Quick Install (Stand-alone Binary)
Alternatively, you can use the installation script to download and verify the stand-alone binary:
```bash
curl -fsSL -o install.sh https://raw.githubusercontent.com/rmedranollamas/research-cli/main/install.sh
# Review the script's contents
less install.sh
chmod +x install.sh
./install.sh
```
### From Source
Ensure you have `uv` installed, then sync the project:
```bash
uv sync
```
## Usage
Set your `RESEARCH_GEMINI_API_KEY` and run a research task:
```bash
export RESEARCH_GEMINI_API_KEY="your-api-key"
research run "How does quantum entanglement work?"
```
If running from source:
```bash
uv run research run "How does quantum entanglement work?"
```
### Key Commands
- **Deep Research**: `research run "query"` - Multi-step, long-running research.
- **Fast Search**: `research search "query"` - Quick, grounded answers using `gemini-2.0-flash`.
- **Status**: `research status ` - Check progress of a background task.
- **Image Generation**: `research generate-image "prompt" -o robot.png` - Generate AI images.
- **List Tasks**: `research list` - View recent task history.
- **Show Task**: `research show ` - Display a specific report from history.
### Context & Options
- **Files**: `--file path/to/file.pdf` - Use local files as context.
- **URLs**: `--url https://example.com` - Use web pages as context.
- **Verbose**: `-v` or `--verbose` - See the agent's real-time "thoughts" and reasoning.
- **Thinking**: `--thinking [minimal|low|medium|high]` - Control reasoning depth (on supported models).
## Configuration
The CLI can be configured via environment variables:
- `RESEARCH_GEMINI_API_KEY`: (Required) Your Google Gemini API key.
- `RESEARCH_MODEL`: Gemini model for research (default: `deep-research-pro-preview-12-2025`).
- `RESEARCH_DB_PATH`: Path to the SQLite history database (default: `~/.research-cli/history.db`).
- `RESEARCH_POLL_INTERVAL`: Maximum interval in seconds for polling interaction status (default: `10`).
- `RESEARCH_MCP_SERVERS`: Comma-separated list of MCP server URLs for extended tool use.
- `GEMINI_API_BASE_URL`: Optional custom base URL for the Gemini API.
## Agent Skill
A [specification-compliant](https://agentskills.io/) agent skill is included in `skills/gemini-research/`. This allows AI agents to learn how to interact with this CLI autonomously.
## Development
- **Linting & Fixing**: `uv run ruff check . --fix`
- **Formatting**: `uv run ruff format .`
- **Testing**: `uv run pytest tests/`