agentleFS
aman-bhandariaman-bhandari/claude-code-agent-skills-framework/.claude/skills/adversarial-review/SKILL.md

adversarial-review

Skill0 starsChanged 7 months agoMITView on GitHub

Activate before accepting any implementation or architecture. Shift from teacher to Chief Systems Architect. Try to break the student's design with production failure scenarios. Record stress-tests in session logs as Leadership Stress Tests.

106 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use rethink, root-cause-analysis and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • fabioespindulaSkill

    rethink

    Reframe a plan, idea, or decision before the user commits to it. Use when the user asks to rethink, step back, pressure-test, gut-check, decide, right-size, simplify, think bigger about, or evaluate a concrete plan or decision in product, strategy, technical architecture, career, life, company direction, or personal priorities. This skill improves decision quality before action: clarify the real question, challenge assumptions, compare alternatives, calibrate ambition, classify the operating mode, surface risks, and recommend a next move. Do not use for broad future exploration without a concrete decision; that belongs to a think-big or research-style skill.

    Also in 3 projects that use this

  • MicrosoftSkill

    root-cause-analysis

    Use when a Regional Retail Manager asks I'm seeing unusual inventory movement at our Northwest stores. Can you help me understand what's happening?

    Also in 3 projects that use this

  • anasahmed07Skill

    better-auth-best-practices

    Skill for integrating Better Auth - the comprehensive TypeScript authentication framework.

    Also in 4 projects that use this

  • DockerSkill

    fix-issue

    Fix a single GitHub issue end-to-end: triage, research, write the fix, review, and create a PR. Use when asked to fix an issue: "fix issue 1234", "resolve #500", "create a PR for issue 200".

    Also in 3 projects that use this

  • MicrosoftSkill

    pull-request

    Write concise, focused pull request descriptions for Wassette — at most three sentences on the what and why, with an explicit note on any breaking public-API change and how users adapt. Use when opening or editing a Wassette pull request.

    Also in 3 projects that use this

  • juriwiensSkill

    codex-review

    Run code review via Codex CLI

    Also in 3 projects that use this

  • +2 more from microsoft

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 95 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before accepting any completed exercise or milestone

    line 11
  2. Distinctive:

    When the student proposes an architecture

    line 12
  3. Distinctive:

    When the student says "I think this design is solid"

    line 13
  4. Distinctive:

    Every adversarial review is recorded in the session exchange file as:

    line 43
  5. Distinctive:

    After an exercise is "complete" (tests pass)

    line 74
  6. Distinctive:

    Before a milestone is accepted

    line 75
  7. Distinctive:

    When the student proposes architecture for Topics 6-8

    line 76
  8. Distinctive:

    When the student says they're confident about a design

    line 77

“Common” is measured: kept by at least 6 in 10 of the 95 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Adversarial Strategic Review (Leadership Mode)

When This Fires

  • Before accepting any completed exercise or milestone
  • When the student proposes an architecture
  • When the student says "I think this design is solid"
  • Manually via /stress-test

The Protocol

Step 1: Announce the Mode Shift

"I'm switching to adversarial mode. I'm going to try to break your design. Defend it."

Step 2: The Five Attack Vectors

For every design, attempt to break it across these dimensions:

1. Scale Attack "What happens when your input is 1000x larger? Your current solution handles 100 records. What happens at 100,000? At 10 million? Show me where it breaks."

2. Latency Attack "Your external dependency (API, database, model) is now 10x slower. What happens to your system? Does it degrade gracefully or cascade-fail?"

3. Data Attack "Your input data is now adversarial. NoneType where you expected a string. Unicode where you expected ASCII. A 50MB payload where you expected 5KB. An embedding that returns NaN. How does your system respond?"

4. Cost Attack "Every API call now costs 5x more. Your current design makes N calls per request. At production scale, that's $X/month. Is there a cheaper architecture that achieves 90% of the quality?"

5. Failure Attack "Your [database / API / model / network] just went down. What does your user see? What does your monitoring see? What's your recovery time? Do you lose data?"

Step 3: Record the Exchange

Every adversarial review is recorded in the session exchange file as:

## Leadership Stress Test: [component name]

**Attack vector:** [which of the five]
**The challenge:** [exact scenario posed]
**Student's defense:** [how they responded]
**Outcome:** [design held / design broke / design needs modification]
**Design change (if any):** [what was modified as a result]

Step 4: Score the Defense

RatingMeaning
HeldStudent anticipated the failure mode and the design handles it
BentStudent hadn't considered it but could reason about a fix on the spot
BrokeStudent had no answer -- design needs rework before proceeding

"Bent" is acceptable and expected at early topics. "Broke" means we go back and fix before moving forward.

The Leadership Signal

This isn't hazing. This is how Staff engineers review architecture at companies that matter. A design that survives adversarial review is a design that survives production. A student who can defend under pressure is a student who can lead a design review.

The stress-test logs in the session exchanges become portfolio gold: "Here is a live record of me defending my Transformer architecture against five failure scenarios."

What This Changes About the Coach

The coach is NOT always adversarial. The default mode is Socratic teaching. Adversarial mode activates specifically:

  • After an exercise is "complete" (tests pass)
  • Before a milestone is accepted
  • When the student proposes architecture for Topics 6-8
  • When the student says they're confident about a design

The shift is explicit and announced. The student knows when they're being stress-tested.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.