Adversarial critique on demand. Returns the strongest case AGAINST a draft, plan, or decision before it ships. Forbids generic mush.
Adversarial critique on demand. Returns the strongest case AGAINST a draft, plan, or decision before it ships. Forbids generic mush.
npx skills add https://github.com/mojwang/macbook-dev-setup --skill skeptic复制此命令并粘贴到 Claude Code 中以安装该技能
| name | skeptic |
| description | Adversarial critique on demand. Returns the strongest case AGAINST a draft, plan, or decision before it ships. Forbids generic mush. |
| user-invocable | true |
| allowed-tools | Read, Bash, Grep |
Get the strongest case against an artifact before you ship it. The skeptic agent argues the opposite position — counter-thesis, concrete failure modes, hidden assumptions — drawn from the artifact itself, vault counterweights, and past decision-records of similar shape.
Dispatches the skeptic agent (via the Task tool with subagent_type: skeptic) with $ARGUMENTS as the artifact.
$ARGUMENTS formats:
vault/career/development/<slug>.md"the proposition to argue against"--pr 88Output is structured: Counter-thesis / Failure modes / Hidden assumptions. Each item is artifact-specific and concrete; the agent prompt explicitly bans "on the other hand," "it depends," and unspecified-mechanism risks.
You decide if the objections land. The agent argues; you arbitrate.
Arguments: $ARGUMENTS
Product development co-pilot. Full lifecycle from idea evaluation through growth, with structured artifacts and evidence gates.
Stress-test ideas, decisions, and exploratory thinking with a curated council of operating wisdom organized into five archetypes (expansionist, first-principles, contrarian, industry-outsider, executor). Three modes — /boardroom explore (Socratic kernel work), /boardroom stress (structured pressure-test), /boardroom debate (multi-voice adversarial). Stress + debate add an anonymous peer-review round and a neutral Chairman synthesis. Dispatches the boardroom orchestrator agent.
Security review checks for shell injection, secrets, file access vulnerabilities, and OWASP top 10. Use when reviewing code changes, PRs, diffs, commits, or editing code that handles user input, auth, or file access.
Bootstrap agentic workflow in a project directory. Supports project types (shell, web) for type-specific skills and hooks.
Commit and PR review checks for conventional commit format, diff size limits, branch verification, and secret detection. Use before creating git commits, pull requests, or when reviewing staged changes.
Capture non-obvious lessons from recent activity. Dispatches the reflector agent to scan commits, edits, and dispatches in scope and propose memory-entry candidates ready to accept.