Cortex Logo
Core

Accelerators

Open source tools for building and securing AI systems.

AI CoreKit

tcoretech/ai-corekit

Simplify the launch of key integrated technology stacks for your AI enabled business. Oneliners for deploying the core tools you will need to scale your idea on your own hosting platforms.

Why it exists

Most experiments die before they start, in the days of setup between an idea and somewhere to try it. Collapsing that to a single command on hosting you already own makes a trial cost an afternoon instead of a procurement cycle, and puts the work on infrastructure with boundaries you control.

Shell

n8n-git

tcoretech/n8n-git

Sync, backup, and organize n8n workflows, credentials, and folder structures with Git. A CLI tool for self-hosted n8n instances.

Why it exists

Automation platforms quietly accumulate business-critical logic inside a database nobody reviews and nobody can roll back. Putting workflows and their structure under version control turns that into ordinary change: diffable, revertable, and answerable months later.

Shell

Autobuilder

tcoretech/autobuilder

GitOps auto-deployment service for AI CoreKit.

Why it exists

If the approved route to an environment is slower than the workaround, people use the workaround. Making a git push the only way anything ships means provenance, review history and rollback come as a side effect of deploying rather than as a gate in front of it.

Shell

Monzo MCP

tcoretech/monzo-mcp

Read-only Monzo banking MCP server for Claude Code — balances, transactions, pots, spending analysis.

Why it exists

Read-only by construction, against genuinely sensitive data. It is a working demonstration that an agent can be given something worth having without being given authority to act: constrain the tool surface and the consequences shrink while the usefulness survives.

Python

Codex Session Handoff

tcoretech/codex-session-handoff

Recover Codex sessions into Claude Code handoffs.

Why it exists

Agentic engineering builds up context that is worth more than the harness holding it, and that context is usually lost when a session ends or a tool changes. Making the working state portable keeps the value with the work rather than with the vendor.

Python