Welcome to ctx|
Context infrastructure for teams building with AI agents.
The context layer for AI agents.
Welcome to ctx|, the shared context system for engineering teams that want AI agents to understand their codebase, standards, decisions, and day-to-day workflow.
ctx| connects your repositories, docs, and engineering tools into an organization-scoped knowledge layer. Agents access that layer through MCP, so Cursor, Claude Code, and other clients can ask grounded questions before they plan, write, or change code.
The result is not another chat surface with a bigger prompt. It is a source of truth that keeps compounding as your team connects code, records decisions, and uses agents in real work.
What's ctx|'s vision?
ctx| exists to make AI agents reliable inside real engineering organizations.
Every team adopting agents runs into the same barrier: the model does not know your architecture, your conventions, your ownership boundaries, or the decision you made last month. Those details live across repositories, ADRs, tickets, docs, and people.
Our vision is to make that organizational context available wherever agents do work. ctx| indexes the sources you connect, builds a living knowledge system, and exposes it through one MCP surface that agents can query as part of their normal workflow.
What we stand for
- Context should be shared - Every agent should work from the same team knowledge, not a one-off prompt or a private local cache.
- Git remains the source of truth - Repositories, instructions, ADRs, and connected docs should feed the knowledge layer without replacing your existing workflow.
- Agents should discover before they decide -
ctx_advisorgives agents a place to ask what is standard, what already exists, and what changed. - Teams should control their data - ctx| is open source, supports self-hosting, and is designed for organizations that care about where context is stored and how it is used.
Where we're headed
We're building the context infrastructure layer for teams that use AI agents as part of software delivery.
That means helping teams:
- connect repositories, docs, and engineering tools into one knowledge system,
- keep architecture and standards visible to agents,
- ground MCP workflows in organization-specific context,
- inspect the graph behind agent answers,
- and run the stack in the environment that fits their security model.
Where to go next
Data flow from repo to graph to MCP.
Who is ctx| for?Teams and roles that get the most value.
MCP setupConnect a repo and MCP client.
Get in touchSales, support, and GitHub.
Want a walkthrough?
If you want to see how ctx| fits your team, get in touch.