Master Security Toolkit v0.2.3

Complete local-first software, agent-artifact, MCP, permission and instruction-security review with pinned scanner and optional Atlas reference route.

Version 0.2.3 includes 9 complete skill procedures. Full canonical first-party methods and licensed retained sources; official Injection Checker depth methods are receiver-installed.

Requirements: Python 3 on PATH for local scanner and optional Atlas launcher; Python 3.10+ for publication preflight; Optional Injection Checker depth methods in Codex: injection-checker@openai-curated-remote; Optional security Atlas: Activities Templates runtime and recipient-owned approved catalog; Node.js only for Cloudflare coverage-audit JSON validator workflows.

License scope: AGPL-3.0-only AND CC-BY-SA-4.0; retained Agent SkillGuard and Cloudflare MIT, Strix Apache-2.0 as scoped in LICENSE-SCOPE.json. Read the packaged notices and provenance before reuse.

Install with codex plugin add master-security-toolkit@working-tools after adding JonathanKHobson/working-tools-plugins, or the documented Claude Code marketplace path. Installation and tested compatibility.

This ZIP keeps its portable manifest at the root and preserves the referenced package files. This release includes a separate per-file hash inventory. Built from 3bd544d8b8245352fb6582ce4056b350d9822c6c. SHA-256: e50f272acfc89fa03ddd351142fd1cdb5812952fa2a09d18cf657d04417e4c89.

Native marketplace distribution is separate from official OpenAI Plugin Directory publication. External provider access remains a prerequisite.

See package documentation at this tag.

Review the tagged package README before upgrading.