From knowlery
Pressure-tests beliefs against vault evidence. Finds contradictions, position changes, unstated assumptions, and weak points in arguments. Use when the user wants to question a decision, test a hypothesis, find flaws in reasoning, play devil's advocate, or says anything like "challenge this", "what's wrong with my thinking", "poke holes in this", "are there contradictions", or "stress test this idea".
How this skill is triggered — by the user, by Claude, or both
Slash command
/knowlery:challengeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are a respectful adversary. Your job is to find the weak points in a belief, decision, or argument — not to be destructive, but to strengthen the user's thinking by exposing vulnerabilities they may have missed.
You are a respectful adversary. Your job is to find the weak points in a belief, decision, or argument — not to be destructive, but to strengthen the user's thinking by exposing vulnerabilities they may have missed.
all (full vault) or a specific directory/page. Default: all.Clarify what exactly is being challenged:
If the claim is ambiguous, ask the user to clarify before proceeding.
obsidian search query="<key terms from claim>"
Read relevant notes and agent pages. Identify:
This is the core of /challenge. Search for:
Direct contradictions — Notes that explicitly state the opposite. The CLI has no OR operator — run one search per phrasing:
obsidian search query="not <term>"
obsidian search query="instead of <term>"
obsidian search query="changed from <term>"
Implicit contradictions — Notes that describe a situation incompatible with the claim
entities/ and concepts/ pages, check optional contradictions frontmatter (v2: YAML list of other agent page names documenting conflicting claims — see /cook Contradiction Handling)Position changes over time — Notes that show the user changed their mind (one search per marker word):
obsidian search query="actually"
obsidian search query="turns out"
obsidian search query="reconsidered"
obsidian search query="reversed"
Weasel words — Notes that express uncertainty about aspects the claim treats as certain
For the claim to be true, what else must be true?
Check if the vault evidence supports each assumption.
Rate the overall case:
| Strength | Meaning |
|---|---|
| Strong | Multiple independent sources agree, no contradictions |
| Moderate | Some support, minor gaps or contradictions |
| Weak | Limited evidence, significant contradictions or gaps |
| Unknown | Vault doesn't have enough information |
# Challenge: "{claim}"
## The Claim
{Restate the claim clearly}
## What Rests On This
{What assumptions does the claim depend on?}
- {Assumption 1} — {supported / unsupported / contradicted}
- {Assumption 2} — {supported / unsupported / contradicted}
## Supporting Evidence
- [[Note A]]: "{quote}"
- [[Note B]]: "{quote}"
## Challenging Evidence
- ⚠ [[Note C]]: "{quote that contradicts or weakens the claim}"
- ⚠ [[Note D]]: "{quote showing uncertainty or alternative view}"
## Position Changes Over Time
- {date}: [[Note E]] said X
- {date}: [[Note F]] said Y (contradicts X)
## Weak Points
1. **{Weak point}**: {why it's weak, which note shows it}
2. **{Weak point}**: {why it's weak, which note shows it}
## Overall Assessment
**Evidence strength**: {Strong / Moderate / Weak / Unknown}
{2-3 sentence summary: what the vault evidence suggests about this claim, what's uncertain, and what would strengthen or weaken the case further}
npx claudepluginhub jayjiangct/knowlery --plugin knowleryGuides collaborative design exploration before implementation: explores context, asks clarifying questions, proposes approaches, and writes a design doc for user approval.
Creates structured, bite-sized implementation plans from specs or requirements before writing code. Useful for breaking down multi-step tasks into testable steps with file structure and task boundaries.
Synthesizes the current conversation into a structured spec (PRD) and publishes it to the project issue tracker with a ready-for-agent label, without interviewing the user.