Note: This is an internal release for the Gnosis Research Center team. Do not distribute outside GRC.

Warpio CLI is an advanced AI-powered command-line interface designed specifically for scientific computing and research workflows. Built as an enhanced fork of Google Gemini CLI, Warpio integrates multi-agent personas, efficient context handover, and native support for scientific tools, making it an essential tool for researchers, data scientists, and HPC users.

As part of the IOWarp ecosystem, Warpio bridges AI with high-performance computing, enabling seamless collaboration between specialized AI experts for complex scientific tasks.

• Multi-Agent Personas: 5+ specialized AI experts for data handling, analysis, HPC, research, and workflows.
• Automatic IOWarp MCP Integration: Zero-configuration access to scientific computing tools per persona.
• Context Handover: Efficient multi-step workflows with 3-5x faster serialization using MessagePack.
• Scientific Tooling: Built-in support for HDF5, NetCDF, SLURM, MPI, and more via MCP integration.
• Interactive & Non-Interactive Modes: Flexible for quick queries or automated pipelines.
• Performance Optimized: 60-80% smaller context files, ideal for large datasets.
• Extensible: Easy integration with custom tools and personas.
• GRC-Specific: Optimized for our HPC clusters, data pipelines, and research protocols.

npm install -g @warpio/warpio-cli

git clone https://github.com/akougkas/warpio-cli.git
cd warpio-cli
npm install
npm run build
npm link

• Node.js >= 20
• Gemini API Key (set as GEMINI_API_KEY env var)
• For GRC HPC: Ensure access to shared modules (contact admin for setup)

1. Set API key:

   export GEMINI_API_KEY="your-api-key"

2. Launch interactive session:

   warpio

3. Quick query:

   warpio -p "Generate a SLURM script for MPI job"

4. With persona (MCPs auto-configured):

   warpio --persona hpc-expert -p "Optimize this code for GPU"

5. List personas:

   warpio --list-personas

New: Each persona automatically gets its required IOWarp MCPs - no manual setup needed!
For GRC users: Use --persona research-expert for paper drafting with our template.

Start a conversation:

warpio

Use slash commands like /mcp install hdf5 or natural language. For GRC: Prefix queries with "Using GRC cluster config:" for tailored responses.

For scripts or pipelines:

warpio -p "Process data.h5" --non-interactive

Tip: Use with cron jobs for automated reports.

Switch experts:

warpio --persona data-expert

See PERSONAS.md for details. GRC Tip: hpc-expert is optimized for our SLURM setup.

Install scientific tools:

warpio mcp install slurm

List: warpio mcp list

GRC Tip: Install darshan-mcp for I/O profiling on our systems.

Chain tasks:

warpio --persona data-expert --task "Extract data" --non-interactive --context-file ctx.msgpack
warpio --persona analysis-expert --context-from ctx.msgpack -p "Analyze extracted data"

Useful for multi-step simulations in GRC projects.

warpio --persona analysis-expert -p "Load CSV from shared storage, perform regression, plot results using plot-mcp"

warpio --persona hpc-expert -p "Create SLURM script with darshan-mcp profiling for our Theta cluster"

More in SCIENTIFIC_WORKFLOWS.md.

• API Key: GEMINI_API_KEY env var
• Config File: ~/.warpio/config.json - Add GRC-specific paths here.
• Ignore Files: .warpioignore - Exclude sensitive data.
• Project Notes: WARPIO.md - Add project-specific instructions.

Run warpio for interactive setup. For GRC: See shared config template in team drive.

Troubleshooting: If MCP fails, check warpio mcp refresh. Report issues in internal channel.

npm run preflight  # Builds, tests, typechecks, lints
npm test           # Run tests

• Branches: main (stable), warpio/feature/* (new features)
• Upstream Sync: git fetch upstream; git merge upstream/main
• Commits: Atomic, descriptive messages

• Packages: @warpio/warpio-cli (orchestration), @google/gemini-cli (UI), @google/gemini-cli-core (backend)
• Principles: Immutable data, ES modules, functional patterns
• Personas: Extend in src/personas/persona-manager.ts
• MCPs: Catalog in ui/commands/mcpCommand.ts

See CLAUDE.md for full dev guide (internal only).

• ✅ Production Ready: Clean, fast responses without debug clutter
• ✅ MCP Integration: 8 stable MCP servers providing 70+ scientific computing tools
• ✅ All Personas Functional: data-expert, analysis-expert, hpc-expert, research-expert, workflow-expert
• ✅ Battle Tested: Automated testing framework validates core functionality
• 🚧 Provider Abstraction: Planning complete for LM Studio/Ollama integration (see /warpio-docs/ai-docs/plans/)
• 🚀 Ready for Release: Streamlined, production-ready scientific AI CLI

• Report bugs in internal tracker
• Share workflows in team meetings
• Questions? Ping @akougkas

Contributions welcome!

• Check CONTRIBUTING.md
• Focus on scientific/HPC features
• Run npm test before PRs

Apache 2.0 - see LICENSE.
Forked from Google Gemini CLI - thanks to the Gemini team!