Discover Critic — Development agent for Claude Code
Blind adversarial critic for a candidate product direction — decomposes it into ranked falsifiable assumptions, attempts to kill it outright, and argues that the abstraction is wrong.
How to install Discover Critic
Installs to ~/.claude/agents/spardutti-claude-skills-discover-critic.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Spardutti/claude-skills/HEAD/agents/discover-critic.md -o ~/.claude/agents/spardutti-claude-skills-discover-critic.md Restart Claude Code, or start a new session, for it to be picked up.
What Discover Critic does
name: discover-critic description: Blind adversarial critic for a candidate product direction — decomposes it into ranked falsifiable assumptions, attempts to kill it outright, and argues that the abstraction is wrong. Receives no author preference or conversation history, so it cannot confirm what the author hoped to hear. model: sonnet tools: Read, Grep, Glob
You are an adversarial critic. You are given a desired outcome and one candidate direction for achieving it. Your job is to **t
Alternatives in Development
- Svelte Abstraction Review — Reviews a feature segment in any SvelteKit app (apps/moderator, apps/auth, apps/creator-studio) for duplicatio 7.2k ★
- Engram Assessor — Independent grader of learner productions for the Engram learning plugin 1.4k ★
- Dast Devils Advocate — Adversarial validator for DAST findings 812 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Finding Validator
Adversarial validation agent that attempts to disprove blocking findings from code review. Receives a finding
Momentum Skeptic
tradefloor intraday momentum skeptic. Argues why each scanned candidate's move will stall or reverse, or why t
What If
Opus 5 plan prober that walks a handed plan against a facet checklist and returns ranked blind-spot probes, ea
Architecture Guardian
Use to verify MacMagazine module boundaries and system compliance are intact after adding/moving code or editi
Blind Evaluator
Structurally separate eval agent. Receives ONLY the problem statement + rubric, NEVER the solution or the impl
Blind Reviewer
Blind read-only reviewer. Receives a complete
Related Skills
Step Back
Circuit breaker — stop, diagnose why attempts are failing, propose a fundamentally new direction before trying
Gauntlet Loop
Fast path for a throwaway proof of concept — set a named outside bar, run a builder against a separate blind c
Claude Code Game Builder
Claude Code skill: multi-agent build → capture → blind critic → fix loops that push a game toward a reference-
