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.
npx skills add ch015/code-pentester --skill verifiergit clone --depth 1 https://github.com/ch015/code-pentesterWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ch015/code-pentester/verifier)<a href="https://agentmods.dev/skills/ch015/code-pentester/verifier"><img src="https://agentmods.dev/badge/skills/ch015/code-pentester/verifier/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ch015/code-pentester/verifier"><img src="https://agentmods.dev/badge/skills/ch015/code-pentester/verifier.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00037 | $0.02603 |
| Opus 5.5 | $0.00015 | $0.01041 |
| Sonnet 5.5 | $0.00007 | $0.00521 |
| Haiku 4.5 | $0.00004 | $0.00260 |
Grade A, and why
ch015-verifier 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 4d ago.
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.
How it starts
The opening of the file, as written. The whole thing — 175 lines — stays where its author put it; the contents beside it link to each section on GitHub.
독립 검증 (Adversarial Verification) Skill
보안 진단 보고서를 독립적으로 검증하여 오탐, 심각도 오류, 누락된 Finding, Finding 간 의존성을 식별합니다.
서비스 개요
| 항목 | 내용 |
|---|---|
| 서비스명 | 독립 검증 (Adversarial Verification) |
| 방법론 | 증거 재검증 + 교차 의존성 분석 + 심각도 민감도 분석 |
| 출력물 | 검증 보고서 (증거 감사 + 의존성 그래프 + 심각도 재평가 + 누락 탐색) |
| 코드 수정 | ❌ 없음 (읽기 전용) |
| 입력 | VA/Pentest/Red Team 보고서 + 대상 소스코드 |
설계 원칙
독립성 원칙
Principles:
Independent_Context: |
"원본 보고서를 생성한 분석과 별도 세션에서 실행한다.
이전 분석의 가정에 오염되지 않은 상태에서 증거를 재검증한다."
Challenger_Stance: |
"모든 판단(위험 판단과 안전 판단 모두)을 의심한다.
보고서의 결론을 수용하지 않고, 증거로부터 독립적으로 결론을 도출한다.
원본 보고서와 동일한 결론에 도달하더라도 그 과정이 증거 기반이어야 한다."
No_Library_Assumptions: |
"라이브러리의 존재, 이름, 일반적 평판으로 보안 속성을 추론하지 않는다.
프로젝트의 실제 설정과 resolved version의 라이브러리 소스/공식 문서로
적용되는 기본 보안 속성까지 확인한다. 옵션 생략만으로 방어 부재를 판단하지 않는다.
버전이나 적용 조건을 확인하지 못하면 unknown과 후속 확인 작업을 남긴다."
Severity_Is_Contextual: |
"심각도는 고정값이 아니라 다른 Finding과의 관계에 의해 변동한다.
Finding A의 해소가 Finding B의 심각도에 영향을 줄 수 있다.
이 의존성을 명시적으로 식별하고 기록한다."
Phase 구조 (v2 — 4-Phase 통합, P1-3)
Phase R0-sealed : Context Binding (봉인) → target/engagement 경로만 수령, VA 보고서 경로는 봉인
Phase R0.5 : Autonomous Discovery → ★ VA 보고서 미열람 상태의 독립 VA 패스 (A1-A8)
manifest 기반 Large Scale 또는 --verify-full이면 풀패스.
일반 엔게이지먼트는 "R0.5-Lite"(차원별 canonical 질문만)
산출물: 02a_verify_autonomous-<round>.md
완료 게이트 통과 전까지 R1 금지 (Invariant I1)
Phase R1-Unified: Evidence & Context Audit ← 기존 R0 + R1 + R1.2 + R1.5 + R1.7 통합
· R0 서브단계: VA 보고서 파싱
· R1 서브단계: Observed/Unverified/Invalidated 재분류
· R1.2 서브단계 (CRITICAL/HIGH 필수, 나머지 20% 샘플): Semantic Taint Re-trace
· R1.5 서브단계: Attack Prerequisites 독립 재평가
· R1.7 서브단계: Asset Value 태그 대조
· R1.8 서브단계: Impact Gate(3 Proofs) 분류 독립 재검증
· 보상 제어 재검증 (compensating-control.md 참조)
Phase R2-Analysis : Dependency + Sensitivity ← 기존 R2 + R3 + R3.5 통합
· R2 서브단계: Finding 간 교차 의존성 그래프
· R3 서브단계: 심각도 민감도 (What-If)
· R3.5 서브단계: Score Recalculation 항상 수행
trigger: "R1-Unified 완료 후 항상 실행"
Phase R4-GapDiff : Autonomous × VA × Raw Ledger 교차 분석
· Matched / VA_Only / Autonomous_Only / Ledger_Only 분류
· Final report omission audit
· Over_Confidence_Gate 판정
산출물: 02b_verify_gap-<round>.md
Phase R5 : Verification Report → 통합 보고서 생성 (규제 영향 참조 포함, 해당 시)
What ships with it
8 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago Changed · +1 lines e73d83a8dd95
- 24d ago Changed · +4 lines 017b16e52729
- 1mo ago Changed · -866 lines 244e71fa3a36
- 1mo ago First seen · 1,036 lines · 37 tokens per session scan A b7223edc3711
ch015-verifier is a skill published in the GitHub repository ch015/code-pentester (4 stars, last pushed 2d ago), licensed MIT. It adds 37 tokens to every session and 2,603 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-08-31.
Other skills, from other repositories
dekko-verify
Sanity-check a suspiciously low or zero call-graph result from dekko (getcallers, getcallees, findusages, impactedtests, unused) before concluding "no callers" or "dead code," or a heritage/throws-provenance result (query supertypes, query subtypes, query throws) that labels something (external) when you expect it to…
dekko-review-context
Give a code-review or PR-description flow a structural head start before reading a diff line by line. Trigger when the user asks for a PR description, a pre-merge summary, or invokes a code-review flow (/code-review, /review, or an equivalent) against a diff in a repo with a .dekko/ directory.
codeweb
Inspect mapped callers and impact before changing code, find existing implementations before adding one, and review structural changes after an edit. Use during coding tasks; Codeweb product planning uses a separate workflow.
analyze
A codebase analysis workflow for existing projects. It explores the code, adds outside research, and produces a Markdown report about architecture, technical debt, security, performance, and possible improvements.
braintrust
An in-session review panel that has several independent agents examine a design, plan, policy, or decision from different critical viewpoints. It reviews the material already discussed in the current conversation and combines the findings.
review-pr
Visualize a pull request on the Cartographer map — see which entities are affected, the blast radius, and the review path. Use when user says "review pr", "review pull request", or "/cartographer review-pr".