michaellady

Mike Skills — Security skill for Claude Code

Security community

Personal Claude Code skills covering the skill lifecycle (scaffold, install, audit, uninstall) plus chat-log meta-analysis.

How to install Mike Skills

This entry records only its repository, not the path inside it, so there is no exact command to give. Open michaellady/mike-skills and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Mike Skills does

Personal Claude Code skills covering the skill lifecycle (scaffold, install, audit, uninstall) plus chat-log meta-analysis.

Alternatives in Security

Pal — Multi-model AI integration — chat, debugging, code review, planning, security audit 11.3k ★
T3mp3st — autonomous red teaming platform; multi-agent offensive-security meta-harness 5.7k ★
Pocketpaw — Your AI agent in 30 seconds 775 ★

README

mike-skills

A workspace of [Claude Code](https://claude.com/claude-code) **skills** — focused, reusable capabilities that Claude loads on demand. Each top-level directory is one skill, defined by a `SKILL.md` (frontmatter + instructions); a few are backed by a small Go binary for the deterministic "transport" parts.

These are the source of truth; the installed copies under `~/.claude/skills/` are symlinks back into this repo (see [Installing](#installing)).

Skills

Multi-AI review & verification

Skill	What it does
converge	Claude, Codex, agy, and the Cursor models Composer 2.5 & Grok Build iterate on an artifact until they converge — or surface a deadlock for you to arbitrate. Five modes: `plan`, `implement`, `verify`, `review`, `audit` (fresh-eyes adversarial review). Backed by a Go binary.
hegelian-dialectic	Works an artifact through an explicit thesis → antithesis → synthesis loop (Claude + Codex) until a transcendent position emerges or the dialectic stalls. Same artifact types as `converge`, different rhythm.
maximize-verification	Stacks every independent check a piece of code admits (differential, property, metamorphic, fuzz, concurrency, static, cross-model) on the strongest available oracle — built to break the correlated-failure trap of one agent writing both code and its tests.
verified-ship	Hard READ-gate state machine for a gated ship pipeline (local verify → commit → push → CI → audit → auto-merge): every gate's real result must be read before advancing, and "run the audit" can never share a turn with "arm the merge." Stops a check being claimed without being read.
probe-before-wire	Before baking an external config value (model id, endpoint, ARN, region, image tag, credential) into code or a deploy, invoke the real dependency once against the target account and read the response — catch a dead