菜单
Use during /setup to test each credential before moving on. Connectivity helpers for Anthropic, Apify, Supabase, Vercel, GitHub, and AssemblyAI.
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
Use when running Module 2B deep niche research. Six pre-built Apify actor recipes parameterized by niche slug. Handles tokens, retries, raw dump storage, rate limits, and cost logging.
Use when writing the website factory brief (Module 5) or content engine brief (Module 7). Owns the two downstream schemas and validates briefs before write.
Use to synthesize raw Apify dumps into the seven (eight including starter template) sub-task markdowns per niche. The Claude-side post-processing layer for Module 2B.
Use when any research-stage agent needs to reference the Student Research System framework. Wraps the framework doc into a queryable phase API so agents don't re-read the whole file.
Use during Module 2D to capture screenshots and DOM data via Apify, score sites with Claude Vision, and scaffold a fresh Vite + React template that the factory can build against.
Use during Stage 4 (asset scraper) of the website factory pipeline to autonomously harvest the client's brand assets, logo, trust badges, project photos, owner/team photos, and any niche-specific asset categories, from their website, Google Business Profile, and linked social channels into the client Assets folder. Halts the entire pipeline if no logo is found. Uses tools/apify-scrape.py for reliable scraping; falls back to native WebFetch only on Apify failure. Never fabricates assets, only flags missing ones.
| name | setup-wizard-toolkit |
| description | Use during /setup to test each credential before moving on. Connectivity helpers for Anthropic, Apify, Supabase, Vercel, GitHub, and AssemblyAI. |
A bag of one-liner tests. Each returns clean pass/fail.
curl -s -o /dev/null -w "%{http_code}" \
-X POST https://api.anthropic.com/v1/messages \
-H "x-api-key: ${ANTHROPIC_API_KEY}" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{"model":"claude-haiku-4-5-20251001","max_tokens":1,"messages":[{"role":"user","content":"hi"}]}'
Expect 200. 401 means bad key. 429 means rate limited (rare during setup).
curl -s "https://api.apify.com/v2/users/me?token=${APIFY_TOKEN}" | jq -r '.data.username // "FAIL"'
Expect a non-empty username. "FAIL" means bad token.
curl -s -o /dev/null -w "%{http_code}" \
"${SUPABASE_URL}/rest/v1/?apikey=${SUPABASE_ANON_KEY}"
Expect 200. 401 means bad anon key. Connection error means bad URL.
Also test service role separately:
curl -s -o /dev/null -w "%{http_code}" \
"${SUPABASE_URL}/rest/v1/?apikey=${SUPABASE_SERVICE_ROLE_KEY}"
Expect 200.
vercel whoami
Expect the student's email or team name. "Error: not authenticated" means they need to run vercel login again.
gh auth status
Expect "Logged in to github.com as {username}". If missing, run gh auth login.
curl -s -o /dev/null -w "%{http_code}" \
-H "authorization: ${ASSEMBLYAI_API_KEY}" \
https://api.assemblyai.com/v2/transcript
Expect not-401 (any other code means key works). 401 means bad key.
Use this idempotent pattern. Don't append duplicate lines:
update_env() {
local key="$1"
local value="$2"
local file=".env.local"
if [ ! -f "$file" ]; then touch "$file"; fi
if grep -q "^${key}=" "$file"; then
sed -i.bak "s|^${key}=.*|${key}=${value}|" "$file" && rm -f "${file}.bak"
else
echo "${key}=${value}" >> "$file"
fi
}
After each successful test, update stack-state.json.credentials.{service}=true using a jq-based update:
jq '.credentials.anthropic = true' stack-state.json > /tmp/state.json && mv /tmp/state.json stack-state.json
Set setup.complete=true. Append a history entry. Tell the student to run /discovery.
Halt at the failing service. Tell the student exactly what failed and what the error means. Do not advance.
Common debug paths:
sk-ant-.vercel logout then vercel login.