v1.1.1

Release Notes: v1.1.1

Prompt Architect v1.1.1 is a packaging and install-hardening release. It keeps the public repository name, folder name, and install target aligned as prompt-architect, and makes the installable skill carry the Prompt Lab knowledge base it references.

Highlights

• Packaged the Prompt Lab KB under .codex/skills/prompt-architect/knowledge/ so SKILL.md references resolve after copying or installing the skill folder.
• Added scripts/install-prompt-architect.ps1 for Windows-first global installation with dry-run, explicit -Yes, backup-before-update, and no delete step.
• Strengthened scripts/validate-prompt-lab.mjs so it verifies root knowledge/ and packaged skill knowledge/ stay byte-for-byte aligned.
• Updated install, validation, skill-structure, README, changelog, and release notes for the packaged KB layout.
• Kept the legacy codex-enterprise-prompt-architect name only as compatibility/history wording; the active repo and skill package name are prompt-architect.

Validation

Run locally:

npm run check
powershell.exe -NoProfile -ExecutionPolicy Bypass -File scripts/install-prompt-architect.ps1 -DryRun
git diff --check
gitleaks detect --redact --no-banner --no-git --verbose

The validator now checks required public files, skill package shape, README assets, local Markdown links, workflow badge honesty, response-mode parity, source-note metadata, generated prompt ledger markers, obvious secret-like values, and packaged KB parity.

Turkish Summary

v1.1.1, Prompt Architect skill kurulumunu sağlamlaştırır. Skill klasörü artık kullandığı Prompt Lab bilgi tabanını da kendi içinde taşır; bu yüzden global kurulumdan sonra knowledge/... referansları boşa düşmez.

Öne çıkanlar:

• .codex/skills/prompt-architect/knowledge/ eklendi.
• Windows için yedekli ve açık onaylı install script'i eklendi.
• Validator, root knowledge/ ile paket içi knowledge/ arasında eksik/fazla/drift dosya kontrolü yapacak şekilde güçlendirildi.
• GitHub repo adı, klasör adı ve skill adı prompt-architect olarak hizalandı.

Known Limitations

• This project is independent and not affiliated with OpenAI.
• It does not enforce runtime security by itself; it provides safer prompt instructions and validation checks.
• It does not publish to npm because the package is intentionally private.
• GitHub repository settings still require manual review in the GitHub UI.

Previous Release

v1.1.0 turned the repository into a validation-backed Prompt Lab package with CI, source discipline, and expanded response modes.

v1.1.0

Release Notes: v1.1.0

Codex Enterprise Prompt Architect v1.1.0 turns the repository from a polished prompt-skill showcase into a validation-backed Prompt Lab package. The release keeps the Codex-first mission intact while adding stronger source discipline, safer tool/subagent routing, and a real local/CI quality gate.

Highlights

• Added dependency-free validation at scripts/validate-prompt-lab.mjs.
• Added GitHub Actions workflow at .github/workflows/docs-validate.yml.
• Added docs/VALIDATION.md, docs/SOURCE_MAINTENANCE.md, docs/SUBAGENTS.md, and docs/PLUGIN_READINESS.md.
• Upgraded README.md and README.tr.md with icon-led command-center navigation and the real validation badge.
• Expanded the skill package with research-backed, template, AGENTS.md, CLAUDE.md, and Gemini system-instruction routes.
• Strengthened MCP, browser, account-tool, and subagent safety guidance.
• Refreshed source notes for current OpenAI/Codex docs, OpenAI tools, OWASP LLM risks, Playwright MCP usage, prompt research, and context-engineering lessons.
• Added .playwright-mcp/ to .gitignore so local browser artifacts do not become public release noise.

Validation

Run locally:

node scripts/validate-prompt-lab.mjs
git diff --check
gitleaks detect --redact --no-banner --no-git --verbose --source .

The validator checks required public files, skill package shape, README assets, local Markdown links, workflow badge honesty, response-mode parity, source-note metadata, generated prompt ledger markers, and obvious secret-like values.

Turkish Summary

v1.1.0, repoyu sadece iyi görünen bir prompt-skill vitrini olmaktan çıkarıp doğrulama kapısı olan bir Prompt Lab paketine taşır.

Öne çıkanlar:

• Lokal validator eklendi.
• GitHub Actions validation workflow eklendi.
• Validation, source maintenance, subagent routing ve plugin-readiness dokümanları eklendi.
• README ve Türkçe README ikonlu komuta merkeziyle daha net hale getirildi.
• Skill routing; research-backed prompt, template, AGENTS.md, CLAUDE.md ve Gemini system instruction üretimini kapsayacak şekilde genişletildi.
• MCP, browser tool, account tool ve subagent güvenlik sınırları güçlendirildi.
• .playwright-mcp/ local artifact klasörü yanlışlıkla stage edilmesin diye ignore edildi.

Known Limitations

• This project is independent and not affiliated with OpenAI.
• It does not enforce runtime security by itself; it provides safer prompt instructions and validation checks.
• It does not install through npm, pip, or a package manager.
• GitHub repository settings still require manual review in the GitHub UI.
• Prompt quality still depends on the user's project context and engineering review.

Previous Release

v1.0.2 improved public repository readiness, bilingual README structure, security documentation, and GitHub contribution routes.

v1.0.2

Release Notes: v1.0.2

Codex Enterprise Prompt Architect v1.0.2 is a public repository readiness release. It keeps the skill package behavior intact while improving the repository storefront, public safety documentation, and GitHub contribution routes.

Highlights

• Reworked README.md into an English-first public repository storefront.
• Updated README.tr.md as the natural Turkish counterpart.
• Added docs/PUBLIC_REPO_CHECKLIST.md for release, publish, and public sharing checks.
• Added docs/SECURITY_MODEL.md for placeholder policy, leak prevention, prompt-injection boundaries, and agent autonomy rules.
• Added documentation and public-safe security report issue templates.
• Replaced private local strategy-note paths in generated prompt outputs with public-safe wording.
• Updated security and contribution docs to point at the new checklist and security model.

Türkçe Özet

v1.0.2, public repo hazırlığı ve güvenlik dokümantasyonu odaklı bir sürümdür. Skill paketinin çalışma davranışı korunmuştur.

Öne çıkanlar:

• README.md global keşif için English-first vitrin olarak düzenlendi.
• README.tr.md doğal Türkçe karşılık olarak güncellendi.
• Release, publish ve public paylaşım öncesi kontrol için docs/PUBLIC_REPO_CHECKLIST.md eklendi.
• Placeholder politikası, sızıntı önleme, prompt injection sınırları ve agent yetki kuralları için docs/SECURITY_MODEL.md eklendi.
• Dokümantasyon ve güvenlik raporu issue template'leri eklendi.
• Üretilmiş prompt kayıtlarındaki private local path ifadeleri public-safe hale getirildi.

Known Limitations

• This project is independent and not affiliated with OpenAI.
• It does not enforce runtime security by itself; it provides safer prompt instructions.
• It does not install through npm, pip, or a package manager.
• Prompt quality still depends on the user's project context and review.

Previous Release

v1.0.1 improved bilingual public documentation, usage examples, FAQ, security wording, and release notes.

v1.0.1

Release Notes: v1.0.1

Codex Enterprise Prompt Architect v1.0.1 is a documentation and public-polish release. It keeps the existing skill package and knowledge base intact while making the repository easier to read in English and Turkish.

Highlights

• English-first README with a clearer first screen.
• Full Turkish README: README.tr.md.
• Bilingual usage, install, examples, FAQ, security, and contribution guidance.
• Cleaner public examples for PROMPT_ONLY, Goal + Full Prompt, no-Goal PLAN MODE ONLY, APPROVED — EXECUTE, STOP / RECOVER, browser QA, and security-sensitive prompts.
• Calmer wording across the public docs and skill package.
• Naming consistency pass for core workflow terms.
• Public safety scan for local paths, credentials, realistic tokens, private keys, and private notes.

Türkçe Özet

v1.0.1, dokümantasyon ve public polish odaklı bir sürümdür. Mevcut skill paketi ve knowledge base korunmuştur.

Öne çıkanlar:

• Daha net, English-first README.
• Tam Türkçe README: README.tr.md.
• Kullanım, kurulum, örnekler, FAQ, güvenlik ve katkı dokümanlarında Türkçe + İngilizce anlatım.
• PROMPT_ONLY, Goal + Full Prompt, no-Goal PLAN MODE ONLY, APPROVED — EXECUTE, STOP / RECOVER, browser QA ve security-sensitive prompt örneklerinde daha temiz anlatım.
• Public repo için daha sakin, daha insan diliyle yazılmış dokümantasyon.
• Local path, credential, gerçek token, private key ve private note taraması.

Known Limitations

• This project is independent and not affiliated with OpenAI.
• It does not enforce runtime security by itself; it provides safer prompt instructions.
• It does not install through npm, pip, or a package manager.
• Prompt quality still depends on the user's project context and review.

Previous Release

v1.0.0 introduced the formal Codex skill package, Prompt Lab knowledge base, response modes, workflow patterns, examples, prompt quality rubric, generated prompt ledger, and public repository documentation.

v1.0.0

Release Notes: v1.0.0

Codex Enterprise Prompt Architect v1.0.0 is the first public release of this Codex-focused prompt-generation skill and Prompt Lab knowledge system.

Highlights

• Formal Codex skill package with SKILL.md entrypoint.
• Reusable command, response-mode, pattern, and example references.
• Goal vs no-Goal guidance for Codex work.
• Strict PLAN MODE ONLY patterns.
• Execute-after-approval prompts.
• STOP / RECOVER prompts for premature execution.
• Browser QA prompt rules for UI work.
• Security-sensitive prompt rules for auth, secrets, password managers, and risky tool use.
• Prompt quality rubric and Codex prompt evaluation workflow.
• Generated prompt ledger.

Included Files

• .codex/skills/codex-enterprise-prompt-architect/SKILL.md
• .codex/skills/codex-enterprise-prompt-architect/commands.md
• .codex/skills/codex-enterprise-prompt-architect/response-modes.md
• .codex/skills/codex-enterprise-prompt-architect/codex-patterns.md
• .codex/skills/codex-enterprise-prompt-architect/examples.md
• knowledge/distilled/
• knowledge/templates/
• knowledge/sources/
• knowledge/outputs/
• docs/

Usage

Ask Codex:

Use the codex-enterprise-prompt-architect skill to create a Codex prompt for: [task]

For prompt-only output:

Yorum yapma, sadece prompt ver: [Codex prompt request]

For strict planning:

Goal kullanmadan plan-only Codex promptu uret: [task]

Known Limitations

• This project is independent and not affiliated with OpenAI.
• It does not enforce runtime security; it generates safer prompt instructions.
• It does not install through npm, pip, or a package manager.
• Prompt quality still depends on the user's project context and review.
• Examples should be adapted before use in high-risk repositories.