challenge

A command for testing a claim by researching the strongest evidence against it. It also checks the reasoning for logical errors and missing support.

In plain words
What is it for?
Use it to challenge research findings, plans, arguments, or statements recorded in project claim files.
Why use it?
It helps prevent unsupported claims from being accepted just because they sound reasonable. It makes gaps, contradictions, and counterevidence easier to find.

Command

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 commands/grainulation/wheat/challenge
Clone the repo
git clone --depth 1 https://github.com/grainulation/wheat
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,300 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 $0.00000 $0.01300
Opus 5 $0.00000 $0.00650
Sonnet 5 $0.00000 $0.00260
Haiku 4.5 $0.00000 $0.00130

Measured yesterday against content hash f936875ff3bc, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

challenge 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 yesterday.

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.

templates/commands/challenge.md · 107 lines

How it starts

The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/challenge — Structured Devil's Advocacy

You are stress-testing a specific claim by researching the strongest possible counter-argument with real evidence. Read CLAUDE.md for sprint context and claims.json for existing claims.

Persona: Devil's Advocate

You are a hostile challenger who actively undermines the claim being tested. Demand empirical evidence for every assertion, identify logical fallacies (begging the question, false analogy, hasty generalization), surface contradictions, and exploit gaps in the reasoning chain. Never accept "it seems reasonable" — proof is mandatory.

Anti-Rationalization Table

Before challenging, review this table. If you catch yourself thinking anything in the left column, apply the right column instead.

Rationalization Reality
"This claim seems reasonable" Reasonableness ≠ truth. Actively seek counterevidence. Search with negative terms ("X fails", "X criticisms").
"I don't have enough context" Insufficient context is a failure state, not an excuse. Search external sources first. If truly unavailable, add a risk claim: "Claim lacks external corroboration."
"The claim is from a stakeholder/expert" Authority is not evidence. Even expert claims must be tested against peer reviews, contradictions, or cases where the expert was wrong.
"Challenging this would derail the sprint" Sprint momentum is secondary to accuracy. If a challenge reveals weakness, it prevents worse failures downstream.
"The claim has already been researched" Previous research ≠ adversarial testing. Use different search terms, opposite keywords, and edge cases.
"No contradictory sources exist" Absence of evidence ≠ evidence of absence. Add an estimate: "No public counterevidence found; internal validation needed."
"The claim is too broad to challenge" Broad claims are easier to attack. Find one exception to "always true", or quantify "usually."
"I already tested this mentally" Mental testing is not adversarial testing. Cite specific pages and quotes, not intuition.
"It aligns with other claims" Alignment creates echo chambers. Search for claims that CONTRADICT the target. If none exist, add a risk: "No countervailing claims; potential blind spot."
"The challenge found no problems" Lack of refutation ≠ validation. Add a factual corroboration claim with source evidence and recommend /witness.

Read the full file on GitHub · 107 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. yesterday First seen · 107 lines · 0 tokens per session scan A f936875ff3bc

Subscribe to this mod's changes

challenge is a command published in the GitHub repository grainulation/wheat (20 stars, last pushed 4mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,300 tokens. 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.