Xirp

What is Xirp?

Xirp is an agentic development environment built around institutional memory. It connects to your services, ownership structures, documentation, and architectural decisions so that every coding session starts with real context—not guesswork. Powered by Spotify Portal, Xirp gives AI agents a working understanding of the system they're operating inside, rather than just the file they're editing.

Who it's for

• Engineering leads who need their AI agents to make decisions that are technically correct and operationally sound, not just syntactically valid.
• Onboarding teams that want to cut ramp-up time from months to days by giving new engineers instant access to accumulated system knowledge.
• Platform engineers managing complex service dependencies who are tired of context living in Slack threads and the heads of long-tenured staff.

Key features

Agent

Xirp's agent doesn't just see the file you're working in—it sees the entire system around it. It knows who owns the upstream service, what depends on it, and why it was built that way. Every session starts grounded in that reality, and each session makes the next one smarter.

Workspace

Workspace is a Portal plugin that holds work items, sessions, and docs in one shared place. When an engineer wraps up a session, the context doesn't vanish. It's stored and ready for the next person or agent who touches that system, creating a continuous thread of understanding.

Documentation

Xirp automatically generates living documentation from every coding session. It captures real knowledge as work happens, keeps it current, and feeds it back into future sessions. No more asking engineers to stop and write docs—the documentation builds itself while the work proceeds.

Flexibility

Xirp works with Claude, Gemini, and Codex, acting as a harness for whichever model you prefer. You can switch between models without losing context, run sessions locally or remotely, and stay vendor-independent while your context layer compounds with every session.

What stands out

"Your agents don't guess. They know."

That's the core difference. Most AI coding tools solve the generation problem—they produce code fast—but they lack the operational awareness that prevents confidently wrong decisions. Xirp solves the retrieval problem instead: the knowledge that would prevent those mistakes already exists in your org, it's just inaccessible. By making that institutional memory available to every agent and engineer in every session, Xirp turns scattered context into a compounding asset.

Worth checking out if…

You're frustrated by AI agents that produce technically correct but operationally wrong code, or if your team's knowledge is trapped in undocumented decisions and hard-to-find threads. Xirp is especially relevant if you're already using Portal or want a development environment where context persists across sessions, models, and team members. It's in beta with access requests open, so it's a good moment to see what your agents can do when they actually know your systems.