Codebase intelligence

Codebase intelligence for AI agents and new teammates

Schemyx maps the pieces agents need to understand: UI components, route structure, API handlers, services, models, theme tokens, and codebase review items. That map can serve agents today and onboarding views tomorrow.

The problem

The codebase graph is usually implicit.

New engineers and AI agents both struggle when knowledge is trapped in folder names, old PRs, and tribal memory. They need to know what connects to what before they make safe changes.

Schemyx approach

Schemyx makes the graph explicit.

The local bundle stores nodes, edges, groups, aliases, terms, recipes, and review items. Agents can fetch the smallest useful context, and teams can eventually visualize the same map for onboarding.

Why teams use it

Less drift, faster onboarding, and the same context across agents.

Outcome

See how product surfaces connect

Routes can point to rendered components. APIs can connect to services and models. Styles can connect to components and UI patterns.

Outcome

Use the same map for onboarding

A new hire can inspect the API layout, database relationships, and UI pattern map instead of asking a teammate to narrate the whole repo.

Outcome

Make uncertainty visible

When several conventions exist, Schemyx records review items so teams can decide what should become the approved path.

Lookup examples

The useful detail lives behind stable keys.

Schemyx is built for exact reads. Agents can ask for a concept, resolve a key, then fetch a tiny context response before they write.

graph/nodes.jsonl

Codebase nodes

Files, components, routes, APIs, services, models, styles, dependencies, and clusters.

graph/edges.jsonl

Codebase edges

Imports, renders, contains, depends_on, calls, styles, reads, and writes relationships.

index/aliases.json

Lookup aliases

Friendly names and tags agents can use before fetching an exact recipe.

review/items.json

Review items

Pattern collisions, duplicate keys, stale approvals, and missing canonical decisions.

Workflow

From codebase scan to agent-safe generation.

Schemyx keeps the heavy context work out of the prompt and turns repeated decisions into reusable lookup responses.

01

Scan code locally

Read only the files inside the workspace scope.

02

Create the graph

Normalize source facts into stable nodes and edges.

03

Resolve intent

Map an agent request to the most useful recipe or cluster.

04

Render maps later

Use the same graph to power visual onboarding, dependency views, and team review flows.

Who it is for

Built for teams and builders who want AI work to stay coherent.

Audience

Growing engineering teams

Help new teammates understand a product faster.

Audience

Teams running agents overnight

Give long-running work a smaller, more reliable context layer.

Audience

Platform-minded founders

Turn codebase knowledge into a reusable asset instead of a recurring explanation.

FAQ

Questions teams ask before trusting agent context.

Is this a visual code map today?

The local bundle already contains graph data. The long-term direction is to expose that graph visually for teams and onboarding.

Can it map backend code too?

Yes. Schemyx can extract API handlers, services, controllers, modules, models, SQL, Prisma, config, and environment hints.

How does this help AI agents?

Agents can retrieve the relevant part of the codebase map instead of searching blind or loading unrelated files.

Does the graph replace architecture docs?

No. It complements docs with current, source-derived structure that can be regenerated as the codebase changes.