Detects prompt injection by its effect on a sacrificial canary model, not just pattern matching: untrusted input hits a powerless model first, a behavioral check reads the residue, and it returns block, flag, or pass before your primary model acts. Inbound preflight sensor, not a guarantee.

Skills for issue-driven software development, from issue to change request.

Portable skills, agents, hooks, and Pi extensions for Claude Code, Codex CLI, Copilot, Cursor, Grok, and Pi.

Multi-agent codebase review for PRs, CI, and downstream fork syncs.

AI-powered development tools. 18 agents, 19 commands, 29 skills, 1 hook, 1 MCP server for code review, research, design, and workflow automation.

AWS の構成図を draw.io の形式で描く Claude Code の Skill / Claude Code Skill that draws AWS architecture diagrams as draw.io files

Project continuity for Claude Code and Codex: a roadmap beside your code, grounded handoffs, and clear task status.

Claude Skill for Backblaze B2 Cloud Storage management from the terminal. Manage buckets, list and search files, audit storage usage, detect stale or large objects, review security posture, and safely clean up data using natural language. Built on the Agent Skills specification and compatible with Claude Code and other skills-based agents.

Claude Code plugin & Agent Skills for AI-native development: codebase readiness scoring (/assess), Six Thinking Hats deliberation (/huddle), AI-slop removal (/deslop), skill hardening (/skill-forge), and more.

The Fintech Builder algorithm library — canonical, cross-language reference implementations of market-data, corporate-action, index, breadth, candlestick and technical-indicator algorithms, exposed as provider-agnostic subpath modules.

Claude Code / Codex / pi plugin that hands agent steps needing no text output to Jev (TypeSafe's judgment model) — measured p50 ~230 ms and ~$0.02 per 1,000 judgments, with typed escalation back to the LLM

Moteur de recherche libre sur le droit français et européen — jurisprudence et textes (loi à n'importe quelle date), en web, API et MCP. Free search engine for French case law and legislation.

SRE thinking applied to Claude Code, based on Boris Cherny's Q&A. It enforces a strict 4-stage pipeline with gates between each step. You literally cannot skip verification. You cannot ship without approval.

Your coding agent designs terrible UI. stitch-kit wires it into Google Stitch MCP and teaches it the whole pipeline — ideation, screen generation, design systems, production components.

A market-driven Product Manager copilot (Agent Skills + Claude Code plugin) based on the Pragmatic Framework — guides PMs from idea to launch and produces a client-ready artifact (Markdown or .docx) at every stage.

SAIL V2 (Secure AI Lifecycle) as an agent skill — the full 91-risk catalog for AI/agent gap assessments, security roadmaps, and compliance checklists. Installs on Claude Code, Codex, ChatGPT, Antigravity, and any SKILL.md-compatible agent.

MetalBear Skills for our users

Claude Code skill that delegates implementation to GPT-5.6 Sol via Codex CLI, Claude plans and reviews the diff, Sol writes the code. Multi-model AI coding where the model that wrote the diff never grades it.

Tech-lead orchestration for Claude Code — the top-tier model (Fable) keeps architecture & decisions, cheap subagents (Sonnet/Opus) do the routine and the digging. Save tokens without losing quality.

One config to rule them all — sync MCP servers, skills and AGENTS.md across 18 AI coding tools, and package them as Agent Plugins.

Football data analytics toolkit for Claude Code. Covers Opta, StatsBomb, Wyscout, SportMonks, and free sources.

A build playbook that turns your coding agent into the foreman of a website build. It interviews you, forces the design and scope decisions, locks a visual system, writes the build brief, then verifies, ships, and indexes the site. Claude Code plugin and Agent Skill. Markdown only, no scripts.

Browse every recent Claude Code session across all your projects and jump back into any of them. Shows project path, summary, git branch, model, turn count, and a ready-to-paste resume command. --open re-opens a whole working set in terminal tabs. Local files only: no network, no telemetry.

Automates mountain route research for North American peaks. Aggregates data from 10+ mountaineering sources to generate detailed route beta reports with weather, avalanche conditions, and trip reports.