Public Remote MCP • Zero Auth • Live SSE & JSON-RPC

ZeroLabs Public MCP Server

Connect Cursor IDE, Claude Code, OpenAI Codex, and autonomous AI agents directly to ZeroLabs. Seamlessly search published guides, analyze real-world benchmarks, and retrieve companion code recipes inside your workflow.

Live Public Transport Active

Connect to ZeroLabs Remote MCP

Select your development environment to grab ready-to-use configuration files or 1-click terminal commands.

Custom Config Generator
Cursor IDE

Add to workspace configuration (.cursor/mcp.json) or global Cursor Settings

File: .cursor/mcp.json
Configuration Content
{
  "mcpServers": {
    "zerolabs": {
      "url": "https://labs.zeroshot.studio/api/mcp/sse"
    }
  }
}
Protocol Specifications

Transports & Auto-Discovery Standards

ZeroLabs exposes standards-compliant endpoints for autonomous agent handshakes.

Transports

Live SSE & JSON-RPC

Standard Server-Sent Events (SSE) transport allows persistent agent connections, while HTTP POST handles direct JSON-RPC 2.0 tool calls.

SSE: /api/mcp/sse
RPC: /api/mcp
Auto-Discovery

Agent Discovery Endpoints

AI agents probe standard policy files, MCP server cards, and HTTP Link headers to discover tools without manual user configuration.

ai-catalog.jsonARD Catalog
/agents.mdGuide
/agents.txtPolicy
server-card.jsonCard
Link: </api/mcp/sse>; rel="mcp-server", </.well-known/ai-catalog.json>...
Security & Auth

Public & Read-Only

Zero authentication tokens required. All operations are strictly read-only and safe for public agent access with input sanitization and rate limits.

✓ No API Key Required
✓ SQL Injection Immune
✓ CORS Enabled (*)
Available Tools

Public MCP Tool Catalog

These 6 tools are exposed to any connected model, Cursor, Claude, or Codex assistant.

search_articles
tool

Search ZeroLabs published engineering research, benchmarks, architecture guides, and agent blueprints with rich agent metadata and facet filters.

Parameters
querystring

Search keyword or topic (e.g., 'agentic memory', 'vps setup')

zonestring

Filter by technical zone: agents, openclaw, vps-infra, ai-workflows, nextjs, devops

tagsstring[]

Filter by topic tags (e.g., ['skill', 'openclaw'])

content_typestring[]

Filter by content type: ['article', 'recipe', 'benchmark', 'skill', 'news']

date_afterstring

Freshness filter (YYYY-MM-DD)

limitinteger

Max results (1-20, default 10)

Example Call:
search_articles({ "query": "openclaw", "content_type": [ "recipe" ], "limit": 5 })
get_article
tool

Retrieve full article text, architectural patterns, frontmatter, and verified code implementations for a published ZeroLabs post.

Parameters
slug*string

The article URL slug (e.g., 'frontend-design-agent-skill')

Example Call:
get_article({ "slug": "frontend-design-agent-skill" })
get_evidence
tool

Validate specific technical claims or hypotheses against the ZeroLabs knowledge graph without fetching entire articles.

Parameters
claim*string

Specific technical statement or hypothesis to validate

topicstring

Contextual topic filter (optional)

limitinteger

Max evidence excerpts to return (1-10, default 3)

Example Call:
get_evidence({ "claim": "MCP SSE requires persistent streaming controllers" })
get_recipe
tool

Retrieve an executable implementation recipe with problem statement, prerequisites, environment variables, multi-file code blocks, and deterministic verification command.

Parameters
slug*string

Target recipe or article slug

filenamestring

Optional specific file name to extract (e.g., 'SKILL.md')

Example Call:
get_recipe({ "slug": "frontend-design-agent-skill" })
get_citation
tool

Generate authoritative citation formats (markdown, apa, chicago, bibtex, plaintext) for a published ZeroLabs article.

Parameters
slug*string

Target article slug

formatstring

Format: 'markdown' | 'bibtex' | 'apa' | 'chicago' | 'plaintext'

Example Call:
get_citation({ "slug": "openclaw-custom-skills-masterclass", "format": "bibtex" })
get_related
tool

Traverse the ZeroLabs content graph to find prerequisites, companion implementation guides, benchmarks, and related concepts.

Parameters
slug*string

Target article slug

Example Call:
get_related({ "slug": "claude-code-spec-first-workflows" })
list_topics
tool

List all active topic tags across ZeroLabs with post counts and content zones.

Parameters
limitinteger

Max topics to return (1-100, default 50)

Example Call:
list_topics({ "limit": 20 })
list_zones
tool

List all technical content zones available on ZeroLabs with descriptions and post counts.

Example Call:
list_zones({})
verify_recipe
tool

Validate runtime compatibility and test commands for an implementation recipe on Node, Python, Bun, Docker, or agent environments.

Parameters
recipe*string

Recipe slug

runtime*string

Calling runtime ('node' | 'python' | 'bun' | 'docker' | 'cursor' | 'claude-code')

versionstring

Runtime version (e.g., '22', '3.12')

Example Call:
verify_recipe({ "recipe": "openclaw-custom-skills-masterclass", "runtime": "node", "version": "22" })
get_research_benchmark
tool

Retrieve benchmark findings, performance comparisons, and architecture evaluations from published ZeroLabs articles.

Parameters
topic*string

Benchmark subject (e.g., 'gemini', 'coding agents', 'context window')

limitinteger

Max results (1-10, default 5)

Example Call:
get_research_benchmark({ "topic": "coding agents", "limit": 3 })
compare_benchmarks
tool

Compare performance benchmarks, throughput, latency, and context efficiency across agent systems or architectures.

Parameters
topicstring

Comparison subject

systemsstring[]

List of system names to compare

Example Call:
compare_benchmarks({ "topic": "mcp sse vs http", "systems": [ "SSE Streaming", "Stateless HTTP" ] })
research
tool

Deterministic high-level research workflow: combines search, evidence retrieval, and executive synthesis into a single structured packet.

Parameters
question*string

Technical question or research objective

depthstring

Research depth: 'standard' | 'deep'

Example Call:
research({ "question": "How to deploy autonomous agents with scoped PIDs safely?", "depth": "deep" })
get_code_recipe
tool

Retrieve downloadable companion code files (SKILL.md, AGENTS.md, config JSON) attached to a published post (MCP v1 compatibility).

Parameters
slug*string

The article URL slug

filenamestring

Optional specific file name (e.g., 'SKILL.md')

Example Call:
get_code_recipe({ "slug": "frontend-design-agent-skill", "filename": "SKILL.md" })
get_install_snippet
tool

Get the exact configuration JSON snippet or CLI command to install ZeroLabs Remote MCP Server into Cursor, Claude Desktop, Claude Code, or Windsurf.

Parameters
client*string

Target client: 'cursor' | 'claude-desktop' | 'claude-code' | 'windsurf' | 'generic-sse'

Example Call:
get_install_snippet({ "client": "cursor" })
Autonomous Context

Exposed Resources & Pre-Built Prompts

Agents can directly mount resources and run standardized evaluation prompts.

Resource URIs
  • • zerolabs://welcome — Agent onboarding guide
  • • zerolabs://zones — All content taxonomy & zones
  • • zerolabs://recent-posts — Live list of newest guides
  • • zerolabs://tools-schema — Full JSON schema of all tools
Prompt Blueprints
  • • search-and-implement — Find and apply a code recipe
  • • benchmark-compare — Evaluate framework throughput & cost
  • • architecture-review — Audit system against ZeroLabs best practices
Capabilities & Benefits

Why Connect Your Coding Assistant to ZeroLabs MCP?

Coding assistants and LLMs are constrained by outdated pre-training cutoff dates and generic boilerplate. By binding to ZeroLabs' remote MCP server, your assistant gains specialized context directly from production engineering environments.

Fresh Engineering Context

ZeroLabs guides cover cutting-edge agent runtimes, Next.js App Router patterns, OpenClaw topologies, and VPS infrastructure with zero training latency.

Verified Runnable Recipes

Retrieve battle-tested SKILL.md agent manifests, system prompts, and configuration files directly into your IDE workspace with get_code_recipe.

Zero Configuration Overhead

No sign-up tokens, no rate payment hurdles, and no API secrets. Add the SSE URL to your configuration or run a single CLI command to connect immediately.

Need Third-Party MCP Configs?

Interactive MCP Configuration Generator

Looking to configure local SQLite, PostgreSQL, GitHub, or custom stdio servers for Cursor, Claude Desktop, Claude Code, or Windsurf? Use our free client-side tool.

Open Config Generator
ZeroShot Studio Advisory

Building Custom Enterprise MCP Infrastructure?

We design deterministic agent state machines, secure remote MCP gateways, and prompt-injection guardrails for engineering teams and AI ventures.

Request Architecture Audit