Web Architecture

Multi-agent orchestration for complex TypeScript/Next.js/Convex projects. Phased builds, functional verification, the full playbook for delegating to sub-agents without chaos.

What to consider before installing or running this skill: - The skill is an instruction-only playbook for coordinating multi-agent development and mostly matches its stated purpose — it doesn't download code or ask for platform installs. - However, the SKILL.md explicitly asks sub-agents to run without time limits. Prefer to keep platform-enforced timeouts or otherwise gate long-running work; do not blindly allow unbounded agent execution. - The workflow expects end-to-end verification (webhooks, payments, auth). Those checks require secrets/API keys (Stripe, Clerk, Convex URL, Resend, etc.). The skill does not declare or request those credentials, so you must not provide production secrets to the agent. If you want it to run E2E tests, create ephemeral/test credentials and run in an isolated/dev environment. - Because the source is unknown and there's no homepage, verify these docs against your codebase and run the skill only on a non-production repository or a sandboxed environment (CI runner, local VM, or ephemeral staging). - Recommended mitigations: supply only test keys, restrict network access for agent runs where possible, keep timeouts enforced at the platform level, and review any outputs before committing or deploying changes the agents produce. Confidence note: assessment is medium confidence — the content is coherent with the described purpose, but the missing explicit credential requirements combined with the 'no timeouts' rule creates practical and safety ambiguities that justify caution.

Type: OpenClaw Skill Name: web-architecture Version: 0.1.0 The skill bundle provides comprehensive documentation and instructions for building a TypeScript/Next.js/Convex application using a multi-agent workflow. All commands and code snippets described are standard development practices (e.g., `bun install`, `npx convex dev`, `next build`). The `SKILL.md` focuses on development methodology and quality assurance, without any prompt injection attempts to subvert the agent's purpose or exfiltrate data. While `TECH-REQ.md` mentions Sentry's session replay with `maskAllText: false`, which could be a privacy concern for the application being built, it is a configuration choice for the application's error tracking, not an indication of malicious intent by the skill bundle itself.