菜单
Launch SEPHIROTH locally and smoke-test the core flow (health, auth, one consultation) — this repo's project verify skill, with this machine's port/venv quirks baked in.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
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。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/Juanpacol/SEPHIROTH --skill verify命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Run the RAG evaluation harness and report metric deltas — use after touching data/rag, the Evidence Agent, or Citation Guard, or whenever the user asks to check/update the eval numbers.
Add a new clinical guideline document to the RAG corpus with matching golden/paraphrase eval cases and tests — use when the user wants to expand data/rag/SEED_GUIDELINES with a new topic.
Run the full local gate (lint, tests+coverage, eval regression check, README/baseline consistency, clean tree) before pushing — use before opening a PR or when the user asks "is this ready to push?".
基于 SOC 职业分类
正在显示 SKILL.md
| name | verify |
| description | Launch SEPHIROTH locally and smoke-test the core flow (health, auth, one consultation) — this repo's project verify skill, with this machine's port/venv quirks baked in. |
This machine runs other Docker stacks that squat the default ports — always use the ports below, not the framework defaults.
| Service | Default | Actual (this machine) |
|---|---|---|
| Postgres | 5432 | 5433 |
| Backend | 8000 | 8000 (but always hit 127.0.0.1, not localhost — IPv4 only) |
| Frontend | 3000 | 3100 |
These are set in .env (gitignored) and read by platform/core/config.py. If .env is missing, recreate it from .env.example and set GEMINI_API_KEY=<your key> and DATABASE_URL=...localhost:5433/....
cd clinical-ai-copilot
# 1. Postgres
docker compose up -d postgres
# 2. Backend (platform/ is on PYTHONPATH, not a package — see CLAUDE.md)
PYTHONPATH=.:platform .venv/bin/uvicorn api.main:app --reload --port 8000 &
curl -s http://127.0.0.1:8000/health
# 3. Frontend
cd platform/frontend && npm run dev -- --port 3100 &
If .venv doesn't exist yet: python3.11 -m venv .venv && .venv/bin/pip install -r requirements.txt (system python3 is 3.9 — too old).
# Health
curl -s http://127.0.0.1:8000/health
# Register + capture token
TOKEN=$(curl -s -X POST http://127.0.0.1:8000/api/auth/register -H "Content-Type: application/json" \
-d '{"email": "verify@test.local", "name": "Verify Bot", "password": "verifypass123"}' \
| python3 -c "import sys, json; print(json.load(sys.stdin)['access_token'])")
# One real consultation through the full multi-agent workflow (calls the Gemini API — burns free-tier quota)
curl -s -X POST http://127.0.0.1:8000/api/agents/consult \
-H "Content-Type: application/json" -H "Authorization: Bearer $TOKEN" \
-d '{"query": "What A1C goal is appropriate for a nonpregnant adult with type 2 diabetes?"}' \
| python3 -m json.tool
Confirm the response has a non-empty answer, citation_report.fabricated == [] (or explains why not), and agents_involved includes "evidence".
GEMINI_API_KEY is set in .env and hasn't hit a quota limit. Error message names the model that failed health check.gemini_rpm_limit) was exceeded — the client retries with backoff automatically; if it persists, wait a minute or lower concurrent consultations.platform/ has no __init__.py at its root (it would shadow the stdlib platform module) and that PYTHONPATH=.:platform is set.