For engineers and their agents

Put your production graph to work.

Xerg connects code, cloud resources and services in a shared graph engineers can explore and agents can query.

WhoEngineers, and their agents
What they ask“What will this change touch?”
Why it mattersPlan changes with evidence
WhoForward Deployed Engineers (FDEs)
What they ask“What does this customer run, what talks to what, and who do I ask for access?”
Why it mattersLearn a customer’s stack without the tour
WhoCoding agents, before every change
What they ask“What does this code connect to?”
Why it mattersGive agents shared context
WhoOn-call engineers
What they ask“What depends on it, who owns it, what changed?”
Why it mattersGet to the cause faster
WhoSREs and platform engineers
What they ask“Which apps depend on Stripe?”
Why it mattersPlan for the next outage
WhoCTO, security and compliance
What they ask“Which outside services do we send data to?”
Why it mattersSupport security reviews
WhoCTO, finance
What they ask“Which resources need a usage or cost review?”
Why it mattersFocus your FinOps review
WhoNew engineers and engineering leaders
What they ask“Where is this used, and how does it all fit?”
Why it mattersGet new people up to speed
An example: Xerg's graph of the public GitHub repo openstatusHQ/openstatus, in 3D. It shows the repo's apps and the databases, cloud resources and outside services they use, and how they all connect, found from the repo's config files and code.
Member of
Agentic AI Foundation
Agentic AI Foundation
FinOps Foundation
FinOps Foundation
The Linux Foundation
The Linux Foundation
NVIDIA Inception ProgramNVIDIA Inception Program
NVIDIA Inception Program
01 / Why now

Your agents can write code. Do they know the system?

Code is only part of the picture. Xerg gives agents and engineers shared context about the systems it connects to, so they can investigate dependencies, review proposed changes and spend less time piecing the same knowledge together.

59%
of developers use AI agents at work, up from 31% in 2025. Stack Overflow, 2026 →
02 / How it works

The graph starts with your code and accounts.

No catalog to build, no YAML to maintain, nothing to install.

01

Connect GitHub.

Xerg reads your config files, import lines and variable names to connect apps, cloud resources and services in a graph you can explore and question.

02

Add account evidence.

Connect AWS, Vercel, Stripe and 16 more. Add account resources and details your code may not reveal. Xerg only reads.

03

Share the context.

Invite your engineers, and anyone else as a free viewer. Give Claude Code or Cursor the same graph to query over MCP.

03 / What it answers

Production knowledge your whole team can use.

Investigate, review and get up to speed.

You found out what depended on that service when it went down.

Plan changes with context

Trace recorded dependencies and source references before a key rotation or vendor switch, then review what needs attention.

ASK · what depends on Stripe?
■ web · checkout, billing portal■ api · /v1/subscriptions■ workers · invoice-sync job3 of 11 items to review
Agent proposals need context beyond the code being edited.

Give agents shared context

Claude Code, Cursor and other agents can query Xerg over MCP to investigate dependencies. Engineers can inspect the same evidence when reviewing proposed changes.

bash — mcp
$claude mcp add --transport http xerg https://mcp.xerg.ai✓ xerg connected · 10 toolsyou › rotate the Stripe keyagent → where_used("STRIPE_SECRET_KEY")api · billing-worker · stripe webhookagent › I found 3 references to review.
Something’s down, and nobody knows who owns it or what changed.

Get to the cause faster

Inspect dependencies, recorded code owners and changes between saved graphs. Xerg keeps a copy of the graph for each day it’s saved, for 180 days.

ASK · what changed in billing this week?
+ Resend · added in web/package.json+ api → Stripe webhook · new routeowners · @payments (CODEOWNERS)
Your SOC 2 vendor list is a spreadsheet nobody trusts.

Support security reviews

Build your system and vendor list for SOC 2 and security questionnaires from repository and account evidence. Security and compliance join free.

ServiceFound inStatus
Stripeapi/package.jsonconnected
AWSinfra/main.tfconnected
Resendweb/.env.exampleconnected
DatadogDatadog accountnot in code
Old integrations and resources can outlast the work they supported.

Find what nobody wrote down

Review account resources with no reference found in scanned code, alongside available 30-day spend for AWS, Azure, OpenAI and Anthropic.

AWS · account resources
lambda · legacy-image-resizes3 · exports-archive2 items to investigate
New engineers take weeks to learn how everything fits together.

Get new people up to speed

New engineers, agents and engineering leaders can explore the same graph, trace dependencies and find the relevant code.

ASK · where is REDIS_URL used?
api · rate limiterworkers · queue consumer→ Upstash · prod, staging
04 / Agents on the team

Agents read the graph. People make the call.

Your agents are teammates now, so Xerg treats them that way. Every agent key is made by a person on your team, and each member’s agents come with their seat.

01 · Live now

Read.

Agents can query the graph and its evidence with 9 tools that never execute external changes: impact, where_used, owners, what_changed and more. Keys allow 600 calls an hour and expire after 90 days.

02 · Next

Act, with approval.

Agents open the pull request or the ticket. A person on your team approves it.

03 · Then

Govern.

Decide, and audit, what each agent can touch and spend.

05 / Trust and price

Read-only by design. Free to start.

Free for one person and their agents. Then $20 per member a month, billed annually, with each member’s agents included. Viewers are always free, you pay only for active members, and no card is needed to start.

See pricing →

Xerg only reads, with the narrowest access each service offers. It never changes anything in your accounts.

✓

No source code or scanned secret values kept. Only variable names, and the service, file and line where something was found.

{}

Know what you connect. Review what Xerg reads, stores and shares before connecting your accounts. Read the security docs →

06 / Questions

Before you connect GitHub.

What is Xerg?

Xerg connects your repositories, cloud resources and services in a shared production graph. Engineers explore it visually, while coding agents query it through MCP. Both can inspect the evidence to understand dependencies and plan changes.

How does Xerg find what we run?

It reads config files, like package manifests, Terraform, deploy configs, CI workflows and example env files, and scans source files for import lines and API hostnames. Connect an account like AWS, Vercel or Stripe to add resources and account details beyond repository discovery.

Does Xerg read our secrets or our code?

The scanner skips real .env files and keeps variable names, never their values. It scans up to 300 source files per repo for imports and API hostnames, and keeps only the service, file and line, never the code. Separately enabled documentation sources retain only the cataloged Markdown, with revision citations. Account connection tokens are stored encrypted.

Can Xerg, or our agents, change anything through it?

Xerg’s connectors and MCP tools never execute external changes. An operator can separately configure a trusted gateway that requires scoped verification before its host runs a fixed command. Installing MCP alone does not enforce that gateway.

How do coding agents use Xerg?

Add Xerg’s MCP server to Claude Code, Cursor or any MCP client with a workspace key. Your agent gets 10 tools to inspect dependencies, retrieve cited documentation, read production evidence and request scoped verification. The tools never execute external changes.

Do agents cost extra?

A member’s agents don’t: they come with the member’s seat, and lookups are never charged per call. An agent with no person behind it, like a CI job, gets its own agent seat, $10 a month billed annually.

How is Xerg different from Backstage or Port?

Xerg starts with a graph from selected repositories and connected accounts, including recorded references to outside services. Engineers and agents query that evidence. Compare coverage, setup effort and the questions each product answers on your own stack.

How much does Xerg cost?

Free for one person and their agents. Team is $20 per member a month billed annually, with each member’s agents included. Viewers are always free.

Is the graph guaranteed to show all of production?

No. Production context records selected sources, serving deployments, observation times and coverage gaps. Optional checks compare selected journeys with docs and bot answers, or inspect runtime and integration evidence. They require configured sources and runners; unconnected systems and unchecked behavior remain unknown.

Does Xerg support GitLab or Bitbucket?

Yes. Hosted scans support GitHub, Bitbucket Cloud and GitLab.com. Self-managed GitLab and Bitbucket Server are not supported.

Put your production graph to work.

Connect GitHub and give your team shared context. Xerg only reads.

Sign up with GitHub Or try it on any public repo, no sign-up →