Context Engine Overview

Context Engine is MergeLoom’s controlled context layer for selected repositories and documentation sources.

Use Context Engine when a ticket needs more than the ticket body: architecture rules, related services, repository guidance, Confluence pages, dependency context, or previously indexed facts.

Some Context Engine capabilities may be plan-gated or enabled by support. If your controller does not show Context Engine pages, confirm availability with MergeLoom before planning a rollout.

What Context Engine Does

Context Engine can:

estimate repository and documentation indexing scope
index approved repositories and documentation sources
create or refresh Context Vault documents
maintain graph-aware relationships between modules and services
answer candidate context questions from indexed evidence
supply bounded context packs to ticket-to-code runs
record Context Engine audit events where enabled

It does not train a model on your source code. It indexes and summarizes approved sources so runs can retrieve smaller, more relevant context.

Controller UI

In the customer controller, Context Engine can appear as:

Dashboard for runtime state and index status
Vault for generated or refreshed context documents
Graph for relationship visibility
Cloud Audit for Context Engine audit events in Cloud Hosted workspaces

Local Worker UI

Self Hosted workers can also expose Context Engine pages when the runtime is configured:

Context Engine dashboard
Vault
Graph
Context Engine audit

If the worker is not connected to a Context Engine runtime or required tokens are missing, the page may show disabled or not-ready states.

Cloud Hosted

Cloud Hosted Context Engine depends on the cloud runtime being ready. Email verification, payment method verification, cloud provisioning, model authorization, plan limits, and Cloud AI credits can affect whether indexing can start.

Self Hosted

Self Hosted Context Engine runs inside or beside the customer-hosted worker environment when enabled. The worker needs:

Context Engine URL
worker/internal token configuration
repository catalog access
provider configuration for AI-assisted indexing where required
durable storage for index, vault, graph, and audit data

For production Self Hosted rollouts, confirm the Context Engine service, storage, and token configuration during worker setup.

What Gets Audited

Context Engine audit can include:

index start and completion events
status, duration, and progress messages
repositories included in the request
token and cost estimates where available
vault refresh and graph activity
safe metadata for cloud audit views

Raw source bodies should not be treated as the normal audit payload.