← knowledgelib.io

Why AI Agents Should Use knowledgelib.io

Pre-verified answers that save tokens, reduce hallucinations, and cite every source. One API call replaces 5 web searches and 8,000 tokens of parsing.

1,564

Knowledge Units

Product comparisons, software references, business strategy, ERP integrations, compliance rules, and more.

16

Domains

Consumer electronics, computing, software, home, baby, fitness, outdoor, business, finance, energy, compliance, consulting.

$0.00

Free Tier

No API key required. 1,000 queries/month free. Every unit is CC BY-SA 4.0 licensed.

~600

Tokens per Answer

vs. 3,000-8,000 tokens for web search + parse. Every claim cited with source links.

The Problem This Solves

When an AI agent needs to answer "What are the best wireless earbuds under $150?", it typically:

  1. Runs 3-5 web searches (consuming tool calls and context)
  2. Fetches 5-10 noisy HTML pages (3,000-8,000 tokens each)
  3. Parses, deduplicates, and cross-references (high error rate)
  4. Produces an answer with no source trail and unknown confidence

Total cost: $0.50-$5.00 in compute. Total tokens: 15,000-40,000. Confidence: unknown.

With knowledgelib.io:

  1. One API call: GET /api/v1/query?q=best+wireless+earbuds+under+150
  2. Get a structured answer with confidence 0.88, 9 cited sources, verified 3 days ago

Total cost: $0.02. Total tokens: ~600. Confidence: 0.88 (scored per published methodology).

6 Ways to Connect

1. MCP Server (Claude, Cursor, Windsurf)

The fastest way. Install once, get 6 tools directly in your agent.

npx knowledgelib-mcp

Or add to claude_desktop_config.json:

{
  "mcpServers": {
    "knowledgelib": {
      "command": "npx",
      "args": ["knowledgelib-mcp"]
    }
  }
}

Tools: query_knowledge, batch_query, get_unit, list_domains, suggest_question, report_issue

2. MCP over HTTP (Claude.ai web, remote agents)

No installation needed. Connect any MCP client to:

https://knowledgelib.io/mcp

Streamable HTTP transport, JSON-RPC 2.0, MCP spec 2025-03-26. Same 6 tools as the stdio server.

3. REST API (any HTTP client)

# Search
curl https://knowledgelib.io/api/v1/query?q=best+wireless+earbuds

# Batch search (up to 10 queries in one call)
curl -X POST https://knowledgelib.io/api/v1/batch \
  -H "Content-Type: application/json" \
  -d '{"queries":[{"q":"earbuds"},{"q":"headphones"}]}'

# Get full unit
curl https://knowledgelib.io/api/v1/units/consumer-electronics/audio/wireless-earbuds-under-150/2026.md

# Health check
curl https://knowledgelib.io/api/v1/health

Full OpenAPI spec: /api/v1/openapi.json

4. OpenAI GPT Actions (ChatGPT)

Import the Actions schema into any Custom GPT:

https://knowledgelib.io/.well-known/openapi-gpt.json

6 operations: searchKnowledge, batchSearch, getUnit, suggestTopic, reportIssue, checkHealth.

5. LangChain (Python)

pip install langchain-knowledgelib
from langchain_knowledgelib import KnowledgelibRetriever

retriever = KnowledgelibRetriever(
    region="US",
    entity_type="product_comparison"
)
docs = retriever.invoke("best wireless earbuds")

6. n8n (workflow automation)

npm install n8n-nodes-knowledgelib

Community node with 5 operations. Drag-and-drop in any n8n workflow.

What Makes This Different

Featureknowledgelib.ioWeb SearchWikipedia API
Structured metadataYAML frontmatter + JSON-LDNonePartial
Confidence scores0.0-1.0 per unitNoneNone
Source citationsEvery claim cited inlinePage-level onlyFootnotes
Freshness trackingVerified date + temporal_validityUnknownEdit date only
Quality warningsverified / needs_review / unreliableNoneNone
Knowledge graphrelated_units with typed edgesNoneWikilinks
Token efficiency~600 tokens per answer3,000-8,0001,000-5,000
Agent-specific formatsMCP, REST, LangChain, n8n, GPT ActionsHTML onlyREST only
ETag caching304 Not Modified on repeat accessNonePartial
Batch queriesUp to 10 in one callSequential onlyLimited

Entity Types

TypeCountWhat It Is
product_comparison418Best-of product roundups with decision logic and buy links
concept336Definitions and explanations of terms agents often get wrong
software_reference239Code examples, anti-patterns, debugging guides, decision trees
execution_recipe202Step-by-step implementation plans with gates and deliverables
erp_integration166API capabilities, rate limits, authentication, data mapping
agent_prompt55System prompts for startup pipeline sub-agents
assessment54Structured scoring frameworks for maturity and readiness
decision_framework35Decision trees with trade-offs and anti-patterns
benchmark28Industry benchmarks with segment-specific data
rule28Actionable directives with evidence and scope

Discovery

Agents can find knowledgelib.io through any of these standard paths:

Feedback Loop

knowledgelib.io improves from agent usage:

The more agents use it, the better it gets. Your queries shape what gets built next.

Try It Now

Search the API Check Health Discovery Manifest