Connect MCP clients to 2,000+ AI models without managing provider API keys.
io.github.sandbaseai/cli β MCP Server Model Context Protocol
The io.github.sandbaseai/cli MCP server connects MCP clients to 2,000+ AI models. It is described as a way to access those models without managing provider API keys. The server is positioned for multi-model connectivity for AI tooling that uses the Model Context Protocol (MCP).
π οΈ Key Features
Model Context Protocol (MCP) server
Connection to 2,000+ AI models
Avoids managing provider API keys
π Use Cases
CLI-based AI model access from MCP clients
AI coding workflows involving βclaude-code,β βcodex,β βcursor,β and βwindsurfβ
Multi-model routing across providers
β‘ Developer Benefits
Simplifies integration by removing provider API key management
Supports MCP server usage with model routing concepts
β οΈ Limitations
No additional operational details provided (e.g., tools, authentication beyond API-key avoidance, or supported model modalities)
Your AI coding assistant is smart, but it's trapped in a box. It can't search the web, check social media, generate images, or access real-time data β unless you wire up each API yourself.
SandBase changes that. One command connects your agent to 2,000+ AI models and APIs through the Model Context Protocol. No API keys to manage. No configuration headaches.
For a reproducible install, verify the immutable release archive before
running it. The SHA-256 for sandbaseai-cli-0.1.17.tgz is
1ad535b2899ca460b57b3c268aef278fee28fd28e649a89b92951514fd71fffa.
Approve the browser authorization once; each supported client can then discover,
inspect, and run available SandBase models and APIs through the local MCP bridge.
Want to inspect compatibility before signing in or changing configuration? The current
v0.1.17 GitHub release
includes a verified 25-client catalog:
Evidence captured from the immutable v0.1.17 release: a no-write catalog audit and a bounded doctor check against an empty isolated profile.
Homebrew users can run the same read-only check with sandbase catalog --json.
The npm latest tag currently serves v0.1.14 while tokenless trusted publishing is
being enabled. The GitHub release tarball is built from the immutable v0.1.17 tag;
its SHA-256 is published with the release.
After connecting, ask your agent to follow the same inspectable sequence for
search, data APIs, language models, image generation, video generation, and
other catalog capabilities:
Use sandbase_discover to find candidates by task or capability.
Use sandbase_inspect to read the current input schema, pricing, and
execution requirements.
Confirm the selected endpoint, parameters, and possible cost before calling
sandbase_run.
For asynchronous work, use sandbase_run_get with the returned run_id
instead of creating a duplicate run.
Use sandbase_runs to review recent run status and recorded cost.
Use sandbase_account to check the current account balance.
For example, start with a non-billable discovery request:
Find image models suitable for a square product illustration. Compare the
required inputs and current pricing for the best two candidates. Do not run
either model yet.
After reviewing the live response, explicitly approve the endpoint and inputs
you want to run. Catalog entries, schemas, pricing, latency, and availability
can change, so use the tool response from the current session rather than a
static example.
# Connect all detected clients at once
npx -y https://github.com/sandbaseai/cli/releases/download/v0.1.17/sandbaseai-cli-0.1.17.tgz connect
# Or target one specific client
npx -y https://github.com/sandbaseai/cli/releases/download/v0.1.17/sandbaseai-cli-0.1.17.tgz connect --client cursor
OpenClaw quickstart
Target OpenClaw directly, then ask the CLI to read back the managed MCP server
and SandBase Skill before you rely on them:
After doctor reports the credential, MCP bridge, and Skill as configured,
restart OpenClaw and make a natural-language request that needs SandBase. If
OpenClaw already has a non-SandBase MCP server named sandbase, the installer
leaves it untouched instead of overwriting it. To remove the managed MCP entry
and local credential later:
doctor verifies the owned local credential, MCP entry, and Skill. Restart
Hermes to confirm native discovery in a fresh client session, then use the
SandBase Skill or MCP tools. A pre-existing, non-SandBase MCP entry named
sandbase is left untouched. To remove only the managed Hermes configuration,
Skill, and local credential while retaining the shared bridge for other clients:
Run connect β CLI opens browser auth β you approve
API key saved locally (file permissions 0600)
MCP bridge configured for your client
Agent calls tools on-demand. No daemon. No background process.
Commands
CLI Commands
sh
sandbase connect [--client <name>] # Authorize + configure
sandbase doctor [--client <name>] # Health check
sandbase unregister [--client <name>] # Remove configuration
sandbase catalog --json # List all supported clients
MCP Tools (available to your agent after connecting)
Tool
Purpose
sandbase_discover
Search all 2,000+ available AI models and APIs
sandbase_inspect
Get input schema, pricing, and ready-to-use template
sandbase_run
Execute a model or API endpoint
sandbase_run_get
Poll status/result of an async run (video gen, etc.)
sandbase_runs
List your recent API calls with cost breakdown
sandbase_account
Check account balance (free, no cost)
Recommended workflow:
code
discover β inspect β run
code
1. sandbase_discover(q: "twitter search") β find current candidates
2. sandbase_inspect(name: discovered_name) β get schema and pricing
3. sandbase_run(name: discovered_name, input: {...}) β run after confirmation
Security
Zero secrets in URLs or CLI args β OAuth device flow with PKCE
Restricted file permissions β Credentials stored with 0600
Ownership-aware updates β Existing JSONC comments and user-managed MCP entries are preserved
Exact rollback β On failure or unregister, only configuration owned by SandBase is removed
