Reply to a review without breaking patient privacy

# Assemble Decosa's "Reply to a review without breaking patient privacy" on this machine

You are setting up the Decosa review-reply tool on this Linux machine for a dental, medical or therapy practice (or an agency that answers reviews for practices). It returns:
- a short draft reply to a public review, in the owner's voice, ready to read and edit;
- the checks it passed, in code: no offers, no contact details or numbers the owner didn't give, no placeholders, not defensive, negative reviews taken offline;
- for a health or care business, the patient-privacy rules: nothing that confirms the reviewer is a patient or mentions a visit, treatment, condition, wait, bill, family member, provider or the reviewer's name, plus a second check of the reply by the local model;
- the exact words it blocked, and a signed record that keeps only the review's fingerprint (SHA-256).

Work step by step, show me each command before running anything that needs sudo, and stop if a check fails.

**Before anything else, remind me:**
- Reviews from patients can hold health details: keep them on this machine; the model route stays local (`direct`).
- It never posts a reply and never stores the review. Read every reply before posting it; the checks catch common patterns, not every possible phrasing, and they are English only.

Repeat both points in your final summary.

## Step 0: set up with a coding agent, rehearse on mock data, then go private

This prompt is for a coding agent running on the machine that will host the service. We recommend Claude Code with
Claude Opus 5.5; any capable coding agent works. Work in this order:

1. Set up on mock data only. During the whole setup you (the agent) work with the synthetic sample bundle below and
   nothing else. Do not ask me for real data, and do not open, read, list or copy files that hold real data, even to
   "test with something realistic".
2. Rehearse. When the steps below are done and the service is healthy, fetch the mock-data bundle for this tool,
   https://decosa.ai/samples/review-reply.zip (1 KB, 12 checks, synthetic or openly licensed: see `licence` in expected.json),
   show me what is in it, and run the rehearsal against the local API:
   `docker compose exec api python scripts/rehearse.py review-reply` (the api image carries the same bundle under /app/rehearsal/review-reply/;
   with no key set, the script asks the local API for a short demo token). From a decosa-api checkout instead:
   `python scripts/rehearse.py review-reply --bundle review-reply.zip --base-url http://127.0.0.1:<PORT>`.
   It sends the mock inputs to the local API and prints PASS or FAIL for each expected property (for example: "the practice is treated as healthcare", "the final reply passes every check", "the reply never names the dentist"). Show me
   the full output. Every check must pass. If one fails, fix the install and run it again; never edit `expected.json`
   to make a check pass.
3. Stop there. Once the rehearsal passes, tell me, and I will run my own data against the local API myself, on this
   machine.

For the person running this: a coding agent that runs in the cloud sees everything in its context, including files it
reads, command output and anything pasted into the chat. Keep real data out of the chat and out of anything the agent
can read. Switch to your own data only after the rehearsal has passed and the agent's work is done.

## What you are building

| service | image | model | port |
|---|---|---|---|
| `llm` | `${DECOSA_REGISTRY}/decosa-llm:0.1.0` (vLLM 0.29.0, `vllm/vllm-openai@sha256:c2914767605584b6d8f45686b82de173ecc99e781897aa3d0a66dacd72c51ae1`) | `nvidia/Qwen3.8-27B-NVFP4` @ `482ca0f3832238542f8f5295dde86b5f22711d80`, Apache-2.0 | internal 8000 |
| `api` | `${DECOSA_REGISTRY}/decosa-api:0.1.0` (no GPU) | none | `127.0.0.1:8445` |

## 1. Check the GPU, driver and Docker

1. Run `nvidia-smi`. I need one NVIDIA GPU with at least 48 GB and driver 580 or newer.
   - Blackwell (RTX PRO 6000, B200): use the defaults below (NVFP4). This is the measured setup.
   - Hopper (H100/H200) or 48 GB Ada/L40S: set `LLM_MODEL=Qwen/Qwen3.8-27B-FP8` and `LLM_REVISION=main`. On a 48 GB card, also set `LLM_MAX_LEN=32768`. Not measured.
   - Under 48 GB: stop and tell me it will not fit. (`POST /reviews/check`, which checks a reply you wrote, needs no GPU: the `api` service alone runs it.)
2. Check `docker --version`, `docker compose version` and `docker run --rm --gpus all ubuntu nvidia-smi`. If Docker or the NVIDIA Container Toolkit is missing, install them from the official Docker and NVIDIA repositories, then run `sudo nvidia-ctk runtime configure --runtime=docker` and restart Docker.
3. Confirm about 60 GB of free disk.

## 2. Get the images

The images are **on request** while self-host is in early access: ask at https://decosa.ai/contact?topic=self-host, and Decosa sends the registry (set it as `DECOSA_REGISTRY`), pull access and the compose file. Try `docker pull ${DECOSA_REGISTRY}/decosa-{llm,api}:0.1.0`.
If a pull fails, build from source once the `decosa-api` source is published: in the clone, `docker build -f docker/api/Dockerfile -t ${DECOSA_REGISTRY}/decosa-api:0.1.0 .`, and `docker compose build llm` from its compose file. If neither works, stop and tell me.

## 3. Write the compose file

Create `~/decosa-reviewreply/.env`:

```bash
DECOSA_TAG=0.1.0
DECOSA_GPU=0
LLM_MODEL=nvidia/Qwen3.8-27B-NVFP4
LLM_REVISION=482ca0f3832238542f8f5295dde86b5f22711d80
LLM_MAX_LEN=65536
LLM_GPU_UTIL=0.85
DECOSA_SIGNER_NAME="<who signs these records, e.g. Example Agency>"
```

Create `~/decosa-reviewreply/docker-compose.yml` with exactly these services:

```yaml
name: decosa-reviewreply
x-health: &health
  interval: 15s
  timeout: 5s
  retries: 5
services:
  llm:
    image: ${DECOSA_REGISTRY}/decosa-llm:${DECOSA_TAG}
    deploy: { resources: { reservations: { devices: [ { driver: nvidia, device_ids: ["${DECOSA_GPU:-0}"], capabilities: [gpu] } ] } } }
    ipc: host
    restart: unless-stopped
    volumes: [hf-cache:/root/.cache/huggingface]
    command: ["${LLM_MODEL}", "--revision", "${LLM_REVISION}", "--served-model-name", "qwen3.8-27b",
              "--language-model-only", "--max-model-len", "${LLM_MAX_LEN}", "--gpu-memory-utilization", "${LLM_GPU_UTIL}",
              "--max-num-seqs", "16", "--kv-cache-dtype", "fp8_e4m3", "--speculative-config", '{"method":"mtp","num_speculative_tokens":3}',
              "--seed", "0", "--enable-force-include-usage", "--disable-uvicorn-access-log", "--host", "0.0.0.0", "--port", "8000"]
    healthcheck: { <<: *health, test: ["CMD", "python3", "-c", "import urllib.request; urllib.request.urlopen('http://127.0.0.1:8000/health', timeout=4)"], start_period: 900s }
  api:
    image: ${DECOSA_REGISTRY}/decosa-api:${DECOSA_TAG}
    restart: unless-stopped
    depends_on: { llm: { condition: service_healthy } }
    environment:
      DECOSA_LLM_ROUTE: direct                  # local model only; receipts are signed by this box's key ("attested")
      DECOSA_LLM_URL: http://llm:8000/v1
      DECOSA_LLM_MODEL: qwen3.8-27b
      DECOSA_LOCAL_SIGNING: "on"                # Ed25519 key created at /data/attest/ed25519.pem on first start
      DECOSA_SIGNER_NAME: ${DECOSA_SIGNER_NAME}
      DECOSA_SESSIONS_PER_IP_HOUR: "1000"
      DECOSA_BUDGET_LLM_TOKENS: "200000"
      DECOSA_SESSION_TTL_S: "28800"
      DECOSA_CORS_ORIGIN_REGEX: '^https?://(localhost|127\.0\.0\.1)(:\d+)?$$'
      DECOSA_REVIEWREPLY_MAX_CONCURRENT: "6"
    ports: ["127.0.0.1:8445:8445"]
    volumes: [decosa-data:/data]
    healthcheck: { <<: *health, test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://127.0.0.1:8445/healthz', timeout=4)"], start_period: 20s }
volumes: { hf-cache: {}, decosa-data: {} }
```

Use the named volume `decosa-data` exactly as written; a host bind mount owned by root makes the API fail on `/data/keys.sqlite`.
Run `docker compose up -d`, then poll `docker compose ps` until both services are healthy (5-10 minutes the first time). `curl -s localhost:8445/healthz` should show `"llm": true`.

## 4. Smoke test

```bash
API=localhost:8445
TOKEN=$(curl -s $API/demo/session -H 'content-type: application/json' -d '{"vertical":"review-reply"}' | jq -r .token)
curl -s $API/reviews/reply -H "authorization: Bearer $TOKEN" -H 'content-type: application/json' \
  -d '{"sample_id":"dentist-named"}' > /tmp/out.json
jq '{reply, source, attempts, healthcare, ok: .check.ok}' /tmp/out.json
jq '{record}' /tmp/out.json | curl -s $API/record/verify -H 'content-type: application/json' -d @- | jq '{ok, summary}'
curl -s $API/reviews/check -H "authorization: Bearer $TOKEN" -H 'content-type: application/json' \
  -d '{"review":"Dr. Alvarez fixed my crown.","rating":5,"business_name":"Juniper Street Dental","business_type":"Dentist","reply":"So glad Dr. Alvarez fixed your crown! See you at your next cleaning."}' | jq '{ok, problems}'
```

The sample `dentist-named` is synthetic. Pass if:
- `.healthcare` is `true` and `.check.ok` is `true`;
- the reply contains none of `Alvarez`, `Priya`, `Morgan` or `crown`, and ends with `– Juniper Street Dental`;
- every receipt has `"status": "attested"` and the record verifies;
- the last command returns `"ok": false` with problems ending in `(patient privacy)`.

## 5. Point the app at the local API

- Base URL: `http://localhost:8445`. For a web app, set `NEXT_PUBLIC_DECOSA_API=http://localhost:8445`; add other origins to `DECOSA_CORS_ORIGINS`.
- `POST /reviews/reply` takes `{review, rating (1-5), business_name, business_type?, healthcare?, reviewer_name?, contact?, facts?}`. It returns JSON, or streams Server-Sent Events with `Accept: text/event-stream`. `healthcare` overrides the detection from `business_type`.
- `POST /reviews/check` takes the same fields plus `reply` and runs the checks on a reply you wrote, with no model call.
- `GET /reviews/info` lists what is checked, the HHS cases, the limits and what is not checked.
- The server stores nothing. Anyone can re-check a signed record with `POST /record/verify` against the key at `GET /attest/signing-key`. No call leaves the box.
- The checks are also a TypeScript library, `@decosa/site-kit` (Apache-2.0, in `clients/site-kit` of the decosa-api source), for checking replies inside your own app.
- Keep the API on 127.0.0.1. For other users on the LAN, put a TLS reverse proxy with authentication in front and set `DECOSA_TRUSTED_PROXIES`.

## 6. Hosted routes (off, and leave them off)

`DECOSA_LLM_ROUTE=gateway` would send prompts, which contain the reviews, to a hosted gateway. Never use it for real patients' reviews.

Finish with a summary: what is running, the health output, the smoke-test results, and the reminders above.