amiteshdwivedijhu-ship-it

Claude Startup Architecture Console — AI skill for Claude Code

AI community

How to install Claude Startup Architecture Console

This entry records only its repository, not the path inside it, so there is no exact command to give. Open amiteshdwivedijhu-ship-it/claude-startup-architecture-console and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Claude Startup Architecture Console does

Unofficial synthetic prototype: Claude Startup Architecture Console for Anthropic's Applied AI Architect, Startups seat (founder discovery -> Claude architecture packet -> first useful agent path -> cited use-case eval -> human gates -> Product/research feedback). No live Claude API, no accounts, not affiliated with Anthropic.

Alternatives in AI

Startup Canvas — Generate Startup Canvas combining Product Strategy and Business Model 7.8k ★
User Research Skill — Cookiy AI Skill for AI agents (Claude, Codex, Cursor, OpenClaw) — end-to-end user research: AI interviews, syn 1.5k ★
Agentty — AI pair programming in your terminal — one static binary, sub-ms startup, any model 597 ★

README

Claude Startup Architecture Console

Working prototype for the Applied AI Architect, Startups seat at Anthropic: a founder-facing working session that turns "we want an agent" into a Claude architecture packet, a first useful agent path, a use-case eval that supports the technical win, and one line of feedback for Product / research.

What it demonstrates

The JD asks for architects who win technical evaluations, develop evals, design architectures, and bring field learnings back to product and research. This console walks a synthetic founder (Maya Chen, founding engineer at Kitepath, a B2B SaaS) through:

Discovery: company stub, agent goal (support triage, draft replies, human approves), constraints
Architecture packet: Messages API + tool use loop, prompt caching, context engineering, human gate before deploy, with an illustrative snippet and citations to public Claude Developer Platform docs
First-deploy checklist: one tool, draft-only week 1, prompt caching, eval set, read-only permissions, budget cap
Use-case eval: compact cited findings (instruction following, tool reliability, long-context cost) from a 20-ticket synthetic eval set
Human gates: Ship first agent path, Tighten tool permissions, Need evals before deploy, Send Product / research feedback
Product / research feedback strip: one field-learning line, sent with one tap

90-second walkthrough

Load: founder + use case stub, architecture packet with 4 chips, deploy checklist (2 open), eval panel with 3 cited findings
Tap "Ship first agent path" → status flips to Ready to ship, checklist completes, eval verdict stands
Reset, tap "Tighten tool permissions" → blocked with reason and missing items
Reset, tap "Need evals before deploy" → auto-reply tool blocked with its own missing eval set
Tap "Send Product / research feedback" → one line logged to the field learnings loop