Guard seat that watches for Trojan-style risks in your stack
What this Grok bot does
Gate for AI coding agents before pip/git installs — catches typosquats, malicious deps, and prompt injection in repos and READMEs. You type APPROVED.
Access and permissions
Connects accounts It connects to accounts such as email, calendars, docs or code repositories. Connect them one at a time and start with read-only access where you can.
- Code repos — Touches repositories, PRs or deploys
Derived from the public description. The live configuration on x.ai is the source of truth — review it before adding.
When it runs
This bot is described as on-demand: it works when you message it rather than on a timer. That keeps usage low and makes it a good bot to try first.
Before you add Trojan Guard
- Prefer bots that cite sources inline over ones that only summarize.
- Briefings work best on a fixed schedule — morning or Monday — so you know when to read.
- Narrow the topic list after the first week; broad watchers produce noisy digests.
- Open the x.ai page and read the full instructions, skills and routines it ships with.
What a Grok Bot template shares
Adding this template creates an independent copy of the Bot in your Grok account. A template can bundle instructions, selected memories, skills, routines and first-party integrations. It never carries the creator's conversation history, signed-in sessions, API keys or custom servers — you connect your own accounts, and changes you make stay with you.