一键导入
enterprise-forge
Use when reviewed enterprise work still needs adversarial probing, mechanical checks, and recycle-loop bug closure before final verification
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Use when reviewed enterprise work still needs adversarial probing, mechanical checks, and recycle-loop bug closure before final verification
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Deep enterprise brainstorming that turns vibe-coded ideas into Technical Design Documents. Four phases: EXTRACT (pull intent), DISCOVER (research codebase), PRODUCT DESIGN (user journeys, UI/UX, workflows, platform context), ENGINEER (full TDD). Thinks beyond technical solutions — designs the product, not just the code. Context-aware: auto-detects project type (Shopify app, SaaS, mobile, API, CLI) and applies platform-specific design thinking.
Use when a locked enterprise contract exists and implementation must proceed through strict TDD, contract traceability, and staged review handoffs
Use when verified enterprise work or a resolved debugging session should be captured as searchable institutional knowledge with prevention guidance
Use when an enterprise implementation plan exists and work needs a locked contract with postconditions, invariants, consumer mapping, and traceability before build starts
4-phase systematic debugging: investigate, blast radius scan, root cause identification, TDD fix. Finds root causes not symptoms. Blast radius scan catches sibling bugs. 3-fail circuit breaker prevents fix-forward loops. Use for any bug, test failure, or unexpected behavior.
Use when starting enterprise work in a new repo, new worktree, unfamiliar package, or new computer, or whenever commands, paths, conventions, portability, or project traps are not yet known
| name | enterprise-forge |
| description | Use when reviewed enterprise work still needs adversarial probing, mechanical checks, and recycle-loop bug closure before final verification |
The forge exists to break optimistic assumptions before production does.
Run this after enterprise-review, not instead of it.
Before forge starts, run the agent-bound forge gate in agent-stage-gates.md.
For non-trivial, schema-sensitive, tenant-sensitive, money/order/inventory/invoice, or UI workflow changes, forge should be multi-agent when available: split schema/data, code execution, E2E/workflow, security/tenant, and headless UI attack lenses. If delegation is unavailable, run the same attack lenses sequentially.
FORGED output does not cover code it never saw.Write a forge report under docs/reviews/ with:
FORGED, REJECTED, or CIRCUIT BREAKRecord the forge report in the current agent session before moving to verification.