翻訳待ち：I pointed my AI security plugin at my own prod app – it refused to invent a bug

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

Notifications You must be signed in to change notification settings Fork 0 Star 0 BranchesTags Open more actions menu Folders and files NameName Last commit message Last commit date Latest commit History 3 Commits 3 Commits .claude-plugin .claude-plugin commands commands skills skills .gitignore .gitignore LICENSE LICENSE PROMPTS.md PROMPTS.md README.md README.md SECURITY.md SECURITY.md Repository files navigation Application security for Claude Code. Audit commands that prove their findings, plus skills that make Claude write secure code before you ask. Built for two people: the engineer who knows what an IDOR is and wants a rigorous second pair of eyes, and the builder who shipped something with AI help and needs a straight answer to "is this safe to put on the internet?" Why this exists Ask any AI to "check my code for security issues" and you get a page of confident, plausible, mostly-wrong findings. Missing headers on a JSON API. SQL injection in a parameterized query. A CVE number that doesn't exist. You read it twice, find nothing real, and never run it again — so the one endpoint that does leak your customers' addresses ships anyway. The problem isn't the model. It's the prompt. securitymaxxing encodes the discipline a real security engineer applies: No finding without a traced path from an attacker-controlled source to a dangerous sink. Both ends named, or it's flagged "needs verification" instead of reported as fact. Check the mitigation first. Most "SQL injection in an ORM" findings die once you read what the framework actually does in that version. Severity by realistic exploitability × impact, not by how frightening the bug class sounds. Zero findings is a valid result. No padding a report to look thorough. Every finding carries the exploit path and the patch — the requests an attacker sends, and the corrected code. Install /plugin marketplace add kenjichristopherv-del/securitymaxxing /plugin install securitymaxxing Or, for local development: /plugin marketplace add ~/Developer/securitymaxxing /plugin install securitymaxxing The commands Type /sec in Claude Code and they'll all filter into view. Command What it does /securitymaxxing:audit Full audit — maps the attack surface first, then works every vulnerability class with traced evidence /securitymaxxing:diff Reviews a diff, branch, or PR. Catches the removed guard clause and the route registered on the wrong router /securitymaxxing:vibe-check Plain-English safety check. No jargon, real rigor, direct verdict /securitymaxxing:ship-check Pre-launch gate. Ten blockers, then should-fixes, then a SHIP / DO NOT SHIP verdict /securitymaxxing:threat-model STRIDE model with a Mermaid data-flow diagram and trust boundaries /securitymaxxing:authz Broken access control — IDOR, mass assignment, tenant isolation. The highest-yield review /securitymaxxing:authn Auth, sessions, JWTs, password reset, OAuth account-linking takeover /securitymaxxing:injection Source-to-sink tracing: SQL, command, XSS, SSRF, path traversal, deserialization, SSTI /securitymaxxing:secrets Sweeps the working tree and git history. Outputs a rotation list, not a cleanup list /securitymaxxing:api CORS, CSRF, headers, rate limits, over-fetching, GraphQL, file uploads /securitymaxxing:crypto Algorithms, IV reuse, key management, TLS verification, what shouldn't be stored at all /securitymaxxing:deps Runs the real auditor, then triages — which of the 140 alerts are actually reachable /securitymaxxing:harden Docker, Kubernetes, Terraform, cloud IAM, and detection readiness /securitymaxxing:llm Prompt injection, tool authority, RAG tenant isolation, unsafe model output /securitymaxxing:redteam Proves it. Builds a working PoC against your own app, and attacks your defenses to confirm they hold /securitymaxxing:fix The only command that writes code. Fixes the class, verifies the patch, adds the regression test /securitymaxxing:explain Explains a vulnerability against your code, so you can spot it yourself next time Every command is read-only except fix (which writes patches) and redteam (which sends active attacks against an app you own). The skills Skills load automatically when Claude detects the relevant work — no command needed. Skill Triggers on secure-coding Writing routes, queries, auth, uploads, shell calls, crypto, or anything taking a resource ID. Supplies the correct pattern first, so the bug never exists llm-app-security Building agents, tool calling, RAG, MCP servers, or any LLM feature security-review-method Any security review — the evidence discipline, severity rubric, and finding format the commands share secure-coding is the quiet one that matters most. It doesn't wait for an audit; it makes the ownership check part of the query the first time the query gets written. Try it /securitymaxxing:vibe-check # start here if you're not sure what you have /securitymaxxing:authz src/api # start here if you are /securitymaxxing:diff # before every commit that touches a route A good workflow: authz and injection on the code you actually shipped, secrets once against full history, redteam to prove the findings are real, diff on every PR, ship-check before launch. Find → prove → fix. The commands come as a loop: a review command surfaces a hypothesis, redteam builds a working proof-of-concept against your own app (and attacks your existing defenses to confirm they hold), and fix closes it with a regression test that keeps the PoC from ever coming back. A finding you've watched get exploited and then blocked is a finding you can trust. Prompt catalog PROMPTS.md — 75+ security prompts as raw, copy-pasteable text. Use them in Cursor, Copilot, ChatGPT, or anywhere else; edit them for your stack; or read them to see exactly what the commands are asking for. Start with Section 0, the preamble. It's the highest-leverage paragraph in the repo — paste it before any security prompt in any tool and the output quality changes immediately. Coverage OWASP Top 10 (2021), OWASP API Security Top 10 (2023), and OWASP Top 10 for LLM Applications, plus the things those lists don't cover well: business logic and race conditions, CI/CD and supply chain, cloud IAM, cost-based denial of service, and incident readiness. What this is not This is a very good first pass and a genuinely good teacher. It is not a penetration test. It reads code — it doesn't run your application, fuzz it, or test it live, so it can't find what only appears at runtime, under load, or in your cloud console. Findings are hypotheses until you verify them. For anything holding real money or real personal data, this belongs alongside a human review, not instead of one. Contributing New prompts, better prompts, and corrections are all welcome — especially real false positives you hit, since those become anti-false-positive rules in security-review-method. Open an issue with the code that got misreported and what the review should have said. License Apache 2.0 — see LICENSE. Resources Readme Apache-2.0 license Security policy Security policy Activity Stars 0 stars Watchers 0 watching Forks 0 forks Report repository