← ClaudeAtlas

dhpk-codex-architectlisted

Codex architecture consulting, including bounded adversarial option convergence. Use when: designing features, evaluating architecture, comparing options, or running independent proposal and critique rounds. Not for: implementation (use dhpk-codex-implement), code review (use dhpk-change-review). Output: architecture advice + design recommendations.
hmj1026/dhpk · ★ 2 · AI & Automation · score 71
Install: claude install-skill hmj1026/dhpk
# Codex Architect Skill (Third Brain) ## When NOT to Use - Code implementation (use /dhpk:dhpk-codex-implement) - Code review (use /codex-review) ## Usage ```bash /dhpk:dhpk-codex-architect "<question>" /dhpk:dhpk-codex-architect "Evaluate this design" --context src/xxx.ts --mode review /dhpk:dhpk-codex-architect "Redis vs MongoDB?" --mode compare /dhpk:dhpk-codex-architect "Which design should we ship?" --mode adversarial ``` ## Modes | Mode | Purpose | When | | ------- | ----------------------- | -------------------------- | | design | Provide design advice | Starting from scratch (default) | | review | Evaluate existing design | Validate solution, find issues | | compare | Compare multiple options | Tech selection | | adversarial | Run independent proposals and bounded critique rounds | Former brainstorm-style exploration with explicit decision criteria | ## Adversarial Mode `--mode adversarial` is the successor route for brainstorm-style architecture exploration. Load [`references/adversarial-option-convergence.md`](references/adversarial-option-convergence.md) before research; it owns the independent-proposal, critique-round, decision-criteria, unresolved-disagreement, and final-recommendation contract. Keep the rounds bounded and report disagreement when the evidence does not support convergence. ## Core Principle ``` User -> Claude -> Codex -> Integrate | | | Initial thinki