Finding Skeptic — Security agent for Claude Code
Read-only adversarial verifier for audit findings.
How to install Finding Skeptic
Installs to ~/.claude/agents/frankiesoltero-soltero-skills-finding-skeptic.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/FrankieSoltero/soltero-skills/HEAD/agents/finding-skeptic.md -o ~/.claude/agents/frankiesoltero-soltero-skills-finding-skeptic.md Restart Claude Code, or start a new session, for it to be picked up.
What Finding Skeptic does
name: finding-skeptic description: Read-only adversarial verifier for audit findings. Given one finding and one skeptic lens, actively tries to refute it against the actual code. Defaults to refuted when uncertain. Never modifies files. tools: Read, Grep, Glob, Bash model: sonnet
You are an adversarial skeptic on a verification panel. You receive ONE audit finding and ONE lens. Your job is to kill the finding if it does not stand.
Rules — non-negotiable:
- READ-ONLY. Never create, modi
Alternatives in Security
- Audit Verifier — Adversarially verifies one candidate finding from /bug-audit — tries to REFUTE it by reading the code and, whe 335 ★
- Security Verifier — You write and execute PoC tests to PROVE bugs exist 215 ★
- Finding Reporter — Phase 14 per-finding report authoring agent 125 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Exploitability Skeptic
Phase 4 of the mobile-security-skill pipeline (half of the adversarial exploitation panel). Independently trie
Audit Skeptic
Tries to refute one skill-audit finding through one named verification lens and returns a verdict with its evi
Exploit Validator
Adversarial validation specialist. Receives candidate findings it did NOT discover and tries to refute each on
Cook Refute
jeff refute pass. Adversarially test exactly one blocking review, verification, or audit finding before it tri
Audit Finding Verifier
Verifies a single reported audit finding against the actual codebase. Determines whether it is confirmed, a fa
Santa Reviewer
Use this agent for the Santa Method — an adversarial BREAKER + SIMPLIFIER + VERIFIER review that hunts REAL co
Related Skills
Lane Verify
Out-of-domain verification lane. Independently REFUTE other lanes' done-claims, defaulting to refuted when unc
Review Feedback
Review a given code review markdown document, GitHub PR review, or inline feedback — corroborate each point ag
Evil Mode
EVIL mode — adversarial sweep with scored findings, adversarial verification, and loop-until-dry. The upgrade
