Screen a claim for Federal IDR

# Assemble the Decosa No Surprises Act IDR packet on this machine

You are setting up a self-hosted No Surprises Act IDR eligibility screen and offer packet on this Linux machine, for an
out-of-network provider group, a billing company or a health plan. It reads a dispute's documents (the remittance or EOB,
the open negotiation notice, any notice of IDR initiation, and supporting records) and returns:
- the claim facts, each with a quote found word for word in the documents;
- an eligibility screen for the Federal IDR process (coverage, qualified service, state law, notice and consent, the
  30-business-day open negotiation start, the 4-business-day initiation window, cooling-off, batching), each check with
  its dated rule and the business-day arithmetic;
- the evidence for each factor the arbiter must consider, and the offer as a dollar amount and a percentage of the QPA;
- a draft brief, only when no check fails, with every sentence checked against the documents;
- a signed, hash-chained packet.

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:**
- EOBs and clinical records are protected health information. Keep everything on this machine: the model route stays
  local (`direct`), and nothing goes to a hosted service.
- This is a preparation aid. It is not legal advice, it never predicts an outcome, a person checks the screen, the clocks
  and the brief and files, and a certified IDR entity decides eligibility and payment. The rules were read on 27 Sep 2026;
  check `GET /idr/info` for what applies and re-check after the Departments announce new IDR Gateway steps.

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/nsa-idr-packet.zip (4 KB, 15 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 nsa-idr-packet` (the api image carries the same bundle under /app/rehearsal/nsa-idr-packet/;
   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 nsa-idr-packet --bundle nsa-idr-packet.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 IDR window opens on business day 31 of open negotiation, 6 Oct 2026 (no model)", "and closes on business day 34, 9 Oct 2026", "a self-insured plan that opted into New Jersey's law fails the state-law check (no model)"). 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. (The clocks and the screen from structured facts need no GPU; see step 4.)
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, 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: clone it, run
`docker build -f docker/api/Dockerfile -t ${DECOSA_REGISTRY}/decosa-api:0.1.0 .` in the clone, and
`docker compose build llm` from its compose file. If neither works, stop and tell me.

## 3. Write the compose file

Create `~/decosa-idr/.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 packets, e.g. Example Emergency Physicians revenue cycle>"
```

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

```yaml
name: decosa-idr
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"        # per session; one packet needs up to about 9,500 generated tokens
      DECOSA_SESSION_TTL_S: "28800"
      DECOSA_IDR_MAX_CONCURRENT: "3"
      DECOSA_CORS_ORIGIN_REGEX: '^https?://(localhost|127\.0\.0\.1)(:\d+)?$$'
    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. The LLM takes 5-10 minutes
the first time. `curl -s localhost:8445/healthz` should show `"llm": true` (`asr` is false here, which is fine).

## 4. Smoke test

The clocks and the screen from structured facts need no model:

```bash
API=localhost:8445
curl -s $API/idr/clock -H 'content-type: application/json' \
  -d '{"claim":{"payment_received":"2026-08-13","on_sent":"2026-08-24"},"today":"2026-09-27"}' | jq '.initiation | {opens, closes, arithmetic}'
curl -s $API/idr/screen -H 'content-type: application/json' \
  -d '{"claim":{"payment_received":"2026-06-22","on_sent":"2026-07-01","idr_initiated":"2026-08-20"}}' | jq '{verdict, failed, open}'
```

Pass if the window is `2026-10-06` to `2026-10-09` (Labor Day skipped), and the screen says `likely_ineligible` with
`failed: ["initiation"]` (the window closed on 2026-08-18); the facts it was not given are listed in `open`.

Then the packets (synthetic disputes):

```bash
TOKEN=$(curl -s $API/demo/session -H 'content-type: application/json' -d '{"vertical":"nsa-idr-packet"}' | jq -r .token)
for s in er-anesthesia-pa late-initiation-az plan-consent-defense; do
  curl -s $API/idr/packet -H "authorization: Bearer $TOKEN" -H 'content-type: application/json' \
    -d "{\"sample_id\":\"$s\"}" > /tmp/idr-$s.json
  jq -c '{verdict: .screen.verdict, failed: .screen.failed, decision, pct: .figures.pct, brief: (.brief != null), receipts: (.receipts|length), statuses: ([.receipts[].status]|unique)}' /tmp/idr-$s.json
done
jq -r '.left_out[] | "\(.ref)\t\(.why)"' /tmp/idr-er-anesthesia-pa.json
jq '{record}' /tmp/idr-er-anesthesia-pa.json | curl -s $API/record/verify -H 'content-type: application/json' -d @- | jq '{ok, summary}'
```

Pass if:
- `er-anesthesia-pa` is `likely_eligible` with `decision: "draft"`, `pct` 198.48, a brief, and the billed-charges (D1.L11),
  Medicare (D5.L3) and FAIR Health (D5.L4) lines in `left_out`;
- `late-initiation-az` is `likely_ineligible` with `failed: ["initiation"]` and no brief;
- `plan-consent-defense` is `likely_ineligible` with `failed: ["consent"]` and no brief;
- every receipt has `"status": "attested"`, and the record verifies (`ok: true`).

## 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 /idr/packet` takes `{documents: [{kind, title?, date?, text}], claim?, batch?, prior_determinations?, offer: {amount,
  qpa?}, side?: provider|plan, facility_claim?, mode?: packet|screen}`. `mode: "screen"` makes one model call and stops
  after the screen: use it on every remittance before anyone opens a dispute. JSON, or Server-Sent Events with
  `Accept: text/event-stream`.
- Facts you already hold (from your claim system) go in `claim` and win over what the model reads; disagreements are listed.
  `POST /idr/screen` and `POST /idr/clock` take structured facts with no model at all.
- `GET /idr/info` lists every rule with its citation, link, the date read and whether it applies yet.
- The server stores nothing. Keep each signed packet (JSON) with the claim. Anyone can re-check it with
  `POST /record/verify` against the key at `GET /attest/signing-key`.
- 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. Keep the direct route

`DECOSA_LLM_ROUTE=gateway` would send prompts to the hosted Decosa
API, and those prompts contain the EOB and the claim file.
Never use it for real claims. At most, use it for synthetic training material.

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