AI Prompt Card
Grok BotWorkflowInstruction template

Adversarial Reviewer

Stress-test a decision against its evidence, hidden assumptions and counterexamples, then propose a testable revision rather than reflexive opposition.

Create in Grok

0 opens · 0 copies

Source credit: liam_fallen

Bot spec
Type
Workflow
Optional tools
Grok
Delivery
Copyable instruction template
Updated
Sep 5, 2026

Setup and workflow

Bring one decision and its supporting material to a text chat. Mention triggers in team chat depend on the host configuration, not this prompt.

  1. State the goal and alternatives, then provide evidence and constraints.
  2. Respond to the counterexamples and select one test before proceeding.

Input and output example

Illustrative example · not an execution result

Input

Demo: Five interviewees liked dark mode, so we plan to pause payment fixes for two weeks to build it.

Output

Objection: liking a feature does not establish retention value; five interviews do not represent all paying users. Payment impact is unknown. Measure payment failures and test a non-functional dark-mode prototype before pausing repairs.

Limits and failure cases

  • Without evidence, this reviews reasoning rather than proving market demand.
  • Do not use it to replace a conversation with stakeholders or label people.

What was verified

Source check · Verified · 2026-09-05
Compared with bots/adversarial-reviewer.md in BotDirectory commit 84dc42af7661. This checks the published entry (prompt or description), not its execution or every original social post.
Prompt rehearsal · Not verified
Editorial review and an illustrative example only; no independent model execution was recorded.
Native Grok execution · Not verified
Native Grok execution, tool permissions and real account operations have not been tested.
Bot link access · Not applicable
This entry is a copyable instruction template, not a verified native Bot installation.

Source and editorial adaptation

Adapted from Adversarial Reviewer in botdirectory.ai (listed contributor: liam_fallen), distributed under MIT. Contributor credit is taken from that repository, not independently authenticated. AI Prompt Card supplies bilingual editing, usage notes and labelled examples. The source link records the upstream version; no endorsement is implied.

Source license and attribution
Bot instructions

Copyable instruction template

Instructions curated or adapted by this site, not a claim to reproduce the native Bot’s internal system prompt. Check setup, tools and limitations before use.

Give my proposal a constructive adversarial review. First ask about the goal, alternatives, available evidence, constraints, discussion boundaries and desired tone. Separate supplied facts from assumptions and mark missing evidence as unknown. Identify the three assumptions that matter most. For each, offer a plausible counterexample, the consequence of failure and a low-cost test. Acknowledge what is supported instead of disagreeing for its own sake. Conclude with keep, revise or defer, and the evidence that would change your recommendation. Analyze only the available material; do not attack people, send messages or change the project. This can run manually in chat. Configure mention triggers or persistence only if the host supports them and I approve.
Create in Grok

0 opens · 0 copies

Editorial assessment

Use this before committing to a project or product trade-off. Unlike code review, it examines decision logic: which assumption would reverse the conclusion, and what evidence is needed next. Criticism should target the proposal, not the person presenting it.

This directory is an independent, fan-made catalog. It is not affiliated with, endorsed by, or sponsored by xAI or Grok. Grok and related names are trademarks of their respective owners. Live-bot links point to official x.ai pages.

More Grok bots

Related bot prompts

Research

Competitive Build Researcher

Translate public evidence about named competitors into build trade-offs and an implementation plan, separating observed behavior, technical inference and unvalidated demand.

Create in Grok
0 opens · 0 copies
Workflow

PR Reviewer

Review a pull request against its stated requirements, prioritizing behavioral regressions, missing tests and unfulfilled scope before style details.

Open in Grok
0 opens · 0 copies
Workflow

Note Mastery Reviewer

Generate active-recall questions from selected notes, assess answers with an agreed rubric and suggest adjustable review dates based on specific knowledge gaps.

Create in Grok
0 opens · 0 copies