Korthread FabricFabric

Context Is All You Need

Your agents reason over fragments scattered across systems, blind to relationships and impact. Korthread℠ Fabric changes that: query a unified graph via MCP or A2A, every token carries more value, every decision costs less, every action lands with full context.

01The Problem

Your Agent Is Only as Smart as Its Context

Status Quo

Fragmented Data

Your agent sees CVE-2024-3271. It queues it as #4,271 in a backlog of thousands.

Better

Partial Picture

Your agent knows the CVE is critical severity. It bumps it up a priority tier.

With
Korthread FabricFabric

Full Context

Your agent sees the CVE affects a PCI-scoped service. The S3 bucket was made public 6 days ago by a Terraform change, the IAM role was expanded last sprint, and together they created a breach path to $2.4M/day in revenue that didn't exist two weeks ago. It escalates to P0 and notifies the VP of Commerce.

The difference isn’t smarter AI. It’s context that connects every signal to business impact.

02Time Machine

Most Platforms Stop at “What Exists”

Knowing what you have is table stakes. The hard questions are: how did you get here, what changed, and who decided? That requires a fundamentally different architecture.

Snapshot

Most tools give you a snapshot of what exists right now. Entities, relationships, current state. Useful for inventory, but silent on how you got here or what’s about to break.

Time Machine

Fabric gives you the full timeline. Who changed what, when, why, and what it triggered downstream. Rewind to any point, trace any decision, see how risk emerged over time.

Fabric is a time machine for your enterprise. Not just what exists, but how things evolved, who changed them, and why it matters now.

03Fabric Skills

Turn Data into Actionable Context

Korthread Fabric doesn't just expose raw graph data. It provides fundamental operational skills that every agent needs, turning queries into actionable intelligence.

  1. 01KDefend
    KDefend
  2. 02KSavr
    KSavr
  3. 03KGovern
    KGovern
  4. 04
    Your Onboarding Agent
    via LangGraph
  5. 05
    Your Incident Agent
    via CrewAI
KDefend
“This CVE became exploitable after Tuesday’s
Korthread FabricFabric
QueryEnrichResolve
IaC:
Owner:
Blast radius:
Works with any agent framework
C
CrewAI
L
LangGraph
L
LangChain
O
OpenAI
A
AutoGen
B
Bedrock
V
Vertex AI
M
Mastra

Raw data is not context. Fabric skills turn context graph queries into operational answers your agents can act on immediately.

04One Query

Graph, Events, and Insights in One Query

Korthread, the engine powering Fabric, captures metadata from your source systems, temporal events and decision traces, and computed insights, all unified by entity ID so your agent gets the full picture in a single query.

Your AgentCrewAILangGraphLangChainOpenAIBedrockAutoGenVertex AIMastra& moreMCPA2AKorthread℠ FabricconsumesCONTEXT AS TOOLSGraph ContextEntities & Relationships from Source SystemsEvent ContextChanges, Traces & Temporal StateInsight ContextEnrichments, Scores & Computed IntelligenceAGENTS AS SKILLSEnvironment IntelligenceDiscovery, Mapping & Risk AssessmentRemediation ExecutionAutomated Response & Workflow ActionsMore agents ...Unified by Entity ID

Graph tells your agent what exists. Events tell it what changed. Insights tell it what it means. Together, they turn every agent action into an informed decision.

05Platform Stack

Three Layers, One Platform

Korthread builds the graph. Fabric makes it queryable and adds operational skills. Agents consume Korthread Fabric to make decisions.

Domain Agents

Domain Agents

Specialized Intelligence

Security, Cost, and Compliance agents that consume Korthread Fabric to deliver autonomous, context-aware decision-making in their domain.

KDefendKSavrKGovernYour Custom Agent
CONSUMES
Korthread Fabric

Korthread Fabric

Service Layer

Exposes Korthread's graph, event history, and decision traces as a queryable service via MCP and A2A, so agents can reason about what changed, why, and what to do next.

Queryable via MCP & A2AOperational SkillsAgent-Ready Intelligence
BUILT ON
Korthread

Korthread

Platform Engine

Continuously ingests from every source, builds a temporally-versioned graph, cross-correlates changes across domains, and surfaces emerging risks as conditions evolve.

Unified ContextAgentic IngestionAgent SDKEnterprise Ready

Agents don’t talk to Korthread directly. Korthread Fabric is the service layer that makes the graph queryable and adds the operational skills every agent needs.

06Integrate

From Zero to Context-Aware in Four Steps

1

Configure Blueprints

Connect your data sources

Connect your data sources using Blueprints. Each one auto-discovers entities, relationships, and metadata, then populates the context graph continuously.

A
AWS
A
Azure
G
GCP
G
GitHub
K
Kubernetes
T
Terraform
O
Okta
S
Snowflake
D
Datadog
J
Jira
S
Salesforce
…
more
2

Authenticate

OAuth2, scopes, and the platform catalog

Obtain an OAuth2 token using client credentials. Optionally request scopes to limit access. Get the MCP server configuration and view registered agents from the platform catalog.

Get Token
OAuth2 client credentials flow
Scopes (optional)
Limit access: context:read, risks:read, events:read
MCP Configuration
Get server URL and settings from the platform
Agent Catalog
View all registered agents, their skills, and permissions
# Get an OAuth2 token
curl -s -X POST "https://auth.<tenant_id>.kscope.ai/oauth2/token" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -u "<client_id>:<client_secret>" \
  -d "grant_type=client_credentials" | jq -r '.access_token'
3

Query Context via MCP or A2A

Use your token to enrich agent reasoning

Pass your token to query context via MCP or A2A. Your agent discovers available tools or skills, then calls them to get rich entity context.

1. DISCOVERY
{
  "mcpServers": {
    "korthread": {
      "url": "https://api.<tenant_id>.kscope.ai/k-mcp/sse",
      "headers": {
        "Authorization": "Bearer <token>"
      }
    }
  }
}
SERVER RESPONDS WITH TOOLS / SKILLS
{
  "tools": [
    {
      "name": "get_entity_context",
      "description": "Get full graph, event, and insight context for any entity",
      "inputSchema": {
        "type": "object",
        "properties": {
          "entity_name": { "type": "string" },
          "include": { "type": "array", "items": { "type": "string" } }
        },
        "required": ["entity_name"]
      }
    },
    {
      "name": "get_compound_risks",
      "description": "Discover compound risks across the context graph",
      "inputSchema": { "..." : "..." }
    }
  ]
}
2. TOOL / SKILL CALL
// Your AI agent automatically calls the tool:
{
  "method": "tools/call",
  "params": {
    "name": "get_entity_context",
    "arguments": {
      "entity_name": "checkout-service",
      "include": ["graph", "insights", "events"]
    }
  }
}
WHAT COMES BACK
{
  "entity": {
    "name": "checkout-service",
    "type": "Application",
    "domain": "Commerce"
  },
  "graph_context": {
    "relationships": [
      { "type": "depends_on", "target": "redis-cluster-prod" },
      { "type": "writes_to", "target": "s3://payment-logs" },
      { "type": "assumes", "target": "iam-role-checkout-admin" },
      { "type": "deployed_via", "target": "ci-pipeline-main" },
      { "type": "owned_by", "target": "checkout-team" }
    ],
    "path_to_revenue": [
      "checkout-service → checkout-team → VP of Commerce → $2.4M/day"
    ]
  },
  "insight_context": {
    "compound_risk_score": 87,
    "business_criticality": "mission_critical",
    "compliance_scope": ["PCI-DSS", "SOC2"],
    "revenue_exposure_daily": 2400000,
    "blast_radius": { "services": 3, "data_stores": 2, "customers": 1200 }
  },
  "event_context": {
    "recent_changes": [
      { "time": "2h ago", "event": "IAM policy expanded", "impact": "blast_radius +2" },
      { "time": "6h ago", "event": "Deployed v3.2.1", "impact": "new dep: log4j-2.14.1" },
      { "time": "2d ago", "event": "S3 bucket → public-read", "impact": "data exposure" }
    ]
  }
}
4

Discover Compound Risks and Act

More value per token, less cost per decision

Individual findings are noise. Your agent threads them across the context graph to discover compound risks: dangerous combinations that no single tool can see. More value per token, lower cost per decision, higher decision quality.

WITHOUT CONTEXT

Agent sees: CVE-2024-3271 (CVSS 7.5)

Agent does: Adds to backlog.
Priority: Medium.
Position in queue: #4,271.

Result: Sits for 90 days.

WITH

Korthread FabricFabric
CONTEXT

Agent sees: CVE-2024-3271 in checkout-service
PCI-DSS scopedexposed S3 bucket (public-read, 2d ago)overprivileged IAM role (expanded 2h ago)deploys to production dailypowers $2.4M/day revenue flowowned by team → VP of Commerce
Compound risk score: 87/100
Agent does: Escalates to P0.
Notifies VP of Commerce.
Creates remediation ticket with blast radius.
Recommends: lock S3, scope IAM, patch CVE.
Result: Resolved in 4 hours.
07See it

Ready to see Korthread Fabric in action?

We use cookies to analyze site traffic and improve your experience. See our Cookie Policy for details.