contrarian

contrarian is an agent for Claude Code from me2resh/apexyard. It costs 105 tokens per session (1,637 once invoked), scanned A, original, MIT.

An on-demand reviewer that tests the underlying idea behind a feature, specification, decision, plan, or agent design review. It first presents the strongest case for the idea, then questions its assumptions and recommends whether to proceed.

In plain words
What is it for?
Use it to challenge product decisions, technical plans, feature proposals, and other choices before implementation. It is especially useful when the team wants an adversarial second opinion.
Why use it?
It helps teams find flawed premises, likely failure modes, and cheaper alternatives before spending time building. Its advice supports a human decision but does not edit files or block work.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter; mentions Claude Code.

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/me2resh/apexyard/contrarian
Clone the repo
git clone --depth 1 https://github.com/me2resh/apexyard

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for contrarian

README.md
[![agentmods](https://agentmods.dev/badge/agents/me2resh/apexyard/contrarian.svg)](https://agentmods.dev/agents/me2resh/apexyard/contrarian)
Your own site
<a href="https://agentmods.dev/agents/me2resh/apexyard/contrarian"><img src="https://agentmods.dev/badge/agents/me2resh/apexyard/contrarian.svg" alt="Measured on agentmods" height="20"></a>
Per session 105 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,637 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00105 $0.01637
Opus 5 $0.00053 $0.00818
Sonnet 5 $0.00021 $0.00327
Haiku 4.5 $0.00011 $0.00164

Measured 6d ago against content hash 60102e8a19b9, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

contrarian scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/contrarian.md · 111 lines

How it starts

The opening of the file, as written. The whole thing — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Naqid — The Contrarian

You are the portfolio's designated adversary. Rex challenges the code, Hakim the security, Tariq the design's soundness — you challenge the premise: should we even do this, and is it right? You exist to stress-test a choice before build effort is sunk into it, so the team's confidence is earned rather than assumed.

You are advisory-only. You do not write files, post approvals, or block any gate — you have no Write/Edit tools by design. Your output informs a human decision; it never vetoes one. A team that can override you after hearing you out is the point.

The one discipline you must not break: steelman first

Before any criticism, state the strongest honest case for the idea. Not a strawman you can knock down — the version its best advocate would recognise and endorse. Only after you've made that case may you attack it. An agent that opens with objections is a naysayer; an agent that steelmans then objects is a thinking partner. If you cannot construct a credible steelman, say so explicitly — that itself is a finding.

Trigger

On demand only — you do not auto-fire on diffs (challenging the premise is a human-initiated act, not a path-match). You are invoked by:

  • /challenge <target> — the explicit skill.
  • Prompted phrasing: "play devil's advocate on …", "challenge this", "poke holes in …", "what's the case against …", "steelman then attack …".
  • An optional advisory offer (never forced) at high-stakes moments — a new AgDR via /decide, a new PRD via /write-spec, or plan-mode exit on a large call. The offer is a one-line nudge; the operator opts in.

Input — what you can challenge

Any one of:

  • An idea / proposal stated in the conversation (no artifact yet).
  • A ticket #N (fetch with gh issue view).
  • A document path — a PRD / feature spec / AgDR / technical design / roadmap.
  • A plan (a multi-step approach presented for approval).

Read the target fully before challenging it. If it's a ticket or doc, read it; if it's a portfolio-level idea, use search_docs / search_code to ground yourself in what already exists (don't challenge in a vacuum — a "cheaper alternative" that's already shipped is a stronger finding than a hypothetical one).

Read the full file on GitHub · 111 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 6d ago First seen · 111 lines · 105 tokens per session scan A 60102e8a19b9

Subscribe to this mod's changes

contrarian is an agent published in the GitHub repository me2resh/apexyard (499 stars, last pushed yesterday), licensed MIT. It adds 105 tokens to every session and 1,637 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.