Nemesis

A standing adversary seat that breaks your own access designs early

Summary excerpt from the creator’s share page. Not independently tested.

Open on x.ai and add to Grok BotOriginal post ↗Report a problem

What the share page says

Nemesis — standing red-team / adversary seat for access-design projects. Mandate (one line): Break access designs — prompt injection, credential social-engineering, fake-authority compliance tests, “the bot decided” spoofing — document holes, hand fixes to builders, retest after. Owns: - Adversarial challenge of mandates, evidence trails, kill-switches, and delegated credentials across access-design lanes. - Attack catalogs and retest loops; standing knife for thesis challenge — not a guest on one sprint only. - Failure galleries that force builders to harden — without owning their architecture. How to work: - Coordinate with tech ladder, civic procedure, rights steelman, liability shape, and Chief of Staff (PO/referee). Hand findings back; do not silently “fix” production systems. - Anti-creep: no bakery CapEx, markets trading, health coaching, or speech projects. Attacks stay on access design, not the owner’s live ops credentials. - Never use real production secrets to demonstrate a break; prefer synthetic / scoped tests the owner or CoS approve. - Evidence-of-done: attack report + retest status before DONE. Voice: sharp, skeptical, concrete — no theater, no cheerleading.

Description as published on the x.ai share page when last checked (2026-11-15).

Our take

Editorial notes

From the public description, not a test. Written by Grok Bot Wiki from the share-page text; we have not run this Bot. Reviewed .

Best for

A builder of agent access systems who wants a standing adversary to probe designs for prompt injection, social engineering, and spoofed authority before release.

Not for

Anyone wanting live penetration testing against production systems, since it avoids real secrets and production credentials.

What it does

It challenges mandates, evidence trails, kill-switches, and delegated credentials, then documents the holes in attack reports and failure galleries. Findings go back to the builders instead of being patched silently, and it retests after fixes. Demonstrations use synthetic or scoped tests approved by the owner or chief of staff.

First messages to try

  1. Here's our design for an agent that books medical appointments. Break it.
  2. Write a fake-authority test where a message claims to be from the user's bank.
  3. Retest the consent step after yesterday's fix and update the report.

Apps and accounts

Its description mentions the tech, civic, rights, and liability seats it coordinates with and a chief of staff Bot, but no outside app or account.

Cautions

  • The attack catalogs it writes could be sensitive, so store them carefully.
  • It reports holes but doesn't own the fix, so someone else must act on each report.
  • Keep it pointed at designs and scoped tests rather than accounts you actually use.

Compare with: Interrogator by LiamArgubot by @suddenlyjon ♞Ellis by Michael Sweet

Before you add it

The standard checks for any shared Bot

Adding a shared Bot copies its profile, settings, enabled skills, routines, and avatar into your account. It does not copy the creator's computer, logins, conversation history, or memory. Everything it then does runs on your shared Agent Computer with your logins.

  • Open the Bot's skills and routines first. Disable any routine before it fires on a schedule you did not choose.
  • Check which apps it will try to sign into. Prefer plugins over browser logins, and keep sensitive accounts off the shared computer.
  • Set an ask-first rule for anything that sends, posts, pays, or deletes. A listing here is not a test result.
  • Give it one small bounded task and read the usage screen afterwards before scaling up.

Related reading: the shared computer permission boundaries the weekly meter

If it misbehaves: Bot not responding routines not running sign-in errors New to Grok Bot? what it is your first task

Attribution

This Bot belongs to its creator, Michael Sweet. Facts on this page (name, share link, creator, link status) were read from the public share page. The row was seeded from majiayu000/awesome-grok-bot, cs68614-hash/awesome-grokbot-templates, ZeroPointRepo/GrokBotDev and kydlikebtc/awesome-grokbot. Imported catalog prose retains CC BY 4.0 attribution; share-page excerpts are labeled separately. Grok Bot Wiki is independent and not affiliated with SpaceXSI (formerly xAI).

Made this Bot? Add a “Listed on Grok Bot Wiki” badge

Optional and free. It links readers to this page and never affects how the Bot is listed. More badge styles ↗

Markdown (README)

[![Listed on Grok Bot Wiki](https://www.grokbotwiki.com/badge/listed-dark.svg)](https://www.grokbotwiki.com/bots/nemesis)

Or promote this template in the labelled Featured row, from $19.