Toolfront provides an MCP server that exposes Statespace data apps—shareable applications built from files and CLI tools—to AI agents for discovery and execution. The server enables agents to read Markdown documentation pages and execute constrained CLI commands (like grep, Python scripts, and SQLite queries) defined within those pages, solving the problem of making local data and tools accessible to AI systems in a controlled, discoverable manner. It bridges the gap between an agent's capability to understand Unix filesystems and the need to safely expose specific data operations and business logic through a standardized interface.
Website: https://statespace.com
Search Markdown and .txt pages from llms.txt sites. Available as a CLI, SDK, MCP server, and agent skill.
npm install @statespace-tech/statespace
Plain queries search across all sites:
mcp server setup
vector database embeddings
rate limiting middleware
oauth2 token refresh
websocket reconnection strategy
Scope a query to a specific site with site: query:
stripe: webhook verification
supabase: edge functions auth
vercel: edge middleware caching
mistral.ai: function calling # match by domain
docs.aws.amazon.com: s3 presigned urls # match by subdomain
Use quotes for exact phrase matching:
"context window limit"
vector database "semantic search"
stripe: "webhook signature verification"
Search documentation from the command line: npx @statespace-tech/statespace search <query> [options]
npx @statespace-tech/statespace search "mcp server setup"
npx @statespace-tech/statespace search "stripe: webhook verification" --limit 5
npx @statespace-tech/statespace search "redis connection pooling" --limit 10 --offset 3
npx @statespace-tech/statespace search "anthropic: tool function calling" --limit 5 --human
| Flag | Short | Default | Max | Description |
|---|---|---|---|---|
--limit <n> | -l | 10 | 50 | Max results to return |
--offset <n> | -o | 0 | — | Results to skip (for pagination) |
--human | — | — | — | Human-readable output instead of JSON |
Import and call search() directly from TypeScript or JavaScript.
import { search } from '@statespace-tech/statespace';
const results = await search("mcp server setup");
const results = await search("stripe: webhook verification", { limit: 5 });
const results = await search("redis connection pooling", { limit: 10, offset: 3 });
Add to your MCP config to expose a search tool to your agents:
{
"mcpServers": {
"statespace": {
"command": "npx",
"args": ["@statespace-tech/statespace", "mcp"]
}
}
}
See SKILL.md
Node.js 18+
MIT
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent