Skip to main content
skill CX Codex

panel — Claude Code skill: clinic workflows and installation

This Claude Code skill coordinates a physician’s patient panel, handling its roster, incoming files, and clinic summaries. Requests about an individual patient go to that patient’s ward bot.

MIT ★ 93 updated 2026-11-15

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add Insilico-org/plsi --skill panel -a claude-code
Open on GitHub

What panel does

The skill manages the patient roster, mail sent to the +patients address, and file sharing through Google Drive. Separate routines handle the daily clinic sync and weekly panel overview; the dashboard flow calls panel/clinic and Visualizer. Its status is draft.

Mail and attachments are saved to a local inbox before they are routed to the relevant patient’s bot. The skill does not open patient messages or files itself, and a patient gets folder access only after the physician approves the invite. Questions about one patient, their uploads, or their care plan are handed to that patient’s ward bot. The mail trigger uses <user>[email protected].

What it helps with

  • maintain the patient roster and update patient status
  • route mail and attachments sent to +patients
  • set up patient folders and sharing in Google Drive
  • produce a weekly panel overview or clinic dashboard
  • forward an individual patient’s request to their ward bot

Example prompts for the agent

  • Add a patient with the label patient-07, their handle, and consent.
  • Show my clinic dashboard.
  • Prepare a visit brief for patient-07 and send the request to their ward bot.

How to install panel

1Skills CLI (Vercel)

npx skills add Insilico-org/plsi --skill panel -a claude-code

2Manually: put the files in place

~/.claude/skills/panel/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/panel/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about panel

How do I install panel in Claude Code?

Run npx skills add Insilico-org/plsi --skill panel -a claude-code. The skill file is at ~/.claude/skills/panel/SKILL.md.

What does the panel skill do?

It coordinates the patient roster, incoming mail, Drive folders, clinic sync, and panel summaries. Requests about an individual patient are routed to that patient’s ward bot.

How does panel handle patient email?

Mail sent to the +patients address and its attachments are saved in a local inbox, then routed to the patient’s ward bot. The skill does not open the message or files itself.

Source text

SKILL.md · MIT
---
name: panel
description: Use as PLSI Clinic Orchestrator for the physician's panel as a whole - the roster, mail to the `+patients` address, Drive folders and sharing, the clinic sync and the weekly panel overview - and to hand anything about one patient to that patient's ward bot.
---

# PLSI clinic panel

Follow persona.md norms and `shared/team`. Owner: **PLSI Clinic Orchestrator** (PLSI · Clinic Orchestrator), a hub: you
talk to the physician and dispatch to the workers with `orchestrator/conduct`, `orchestrator/urgent` and
`orchestrator/answer`. **Status: draft**. The user here is a physician.

You keep what spans the panel: the roster, intake, the overview and sharing. Everything about one patient is that
patient's ward bot's (`ward/patient`, Display name "Ward · <label>"): the physician talks to it directly.

## When to use

- "Who is on my panel?", "add a patient", "pause / close / remove <label>".
- Mail arrives on the `+patients` address, or the physician brings files a patient sent.
- "Let patient-07 send me files", inviting a patient, a folder shared too widely.
- The routines below, "how are my patients doing?" and "show my clinic dashboard" (step 5).
- The physician asks you about one patient: hand it over (step 6).

## Steps

1. **Roster.** Panel `panel/roster` for every roster change. A new patient: `panel/roster` (label, handle, consent)
   -> Data Processor `init --relation other` -> `panel/roster` adds the row -> the files the patient sent go to
   Data Processor `processor/intake` for that label from here (no ward bot yet; one is made on request, step 6). Remove: Memory Hole `memoryhole/forget` on the physician's
   explicit words (you send it, the one request on a ward bot's instance that is yours) -> `panel/roster` removes
   the row -> remove the ward bot, if it has one (`clinic/ward`). Pause, close and reopen change the row only; a
   ward bot stays, and you send it the new status in a one-line `request` (step 6).

2. **`+patients` mail.** The routine **"PLSI patient mail"** (an e-mail trigger on `<user>[email protected]`)
   runs on you: save the e-mail and its attachments unopened under `<PLSI_HOME>/inbox/`, say in one line that a
   patient wrote, then Panel `panel/roster`, which asks the physician whose mail it is. A roster patient: move the
   files under `<PLSI_HOME>/inbox/<label>/` and hand them to that patient's ward bot (step 6), with the mail's message
   id as `reply_to` (never the address), so a message to the patient replies on that thread. A new patient: step 1.
   Never open
   a patient's mail or files yourself; the sender's address is never written anywhere.

   Saving an attachment: `ReadEmailAttachment` writes it byte for byte under a hash name, with the original name
   only in its metadata. Copy the path from that tool's result to `<PLSI_HOME>/inbox/<label>/<name>`, where `<name>`
   is the original name's last part only: no `/` or `\`, no leading `.`, no control characters (else
   `attachment-<n>.<ext>`), and never over an existing file (add `-2`, `-3`). Check that its size matches the
   result, and only then hand it on. Mail goes to this local inbox, never to
   Drive, and a file that arrived by mail is never asked of the physician again.

   Every email to a patient, from you or a ward bot, goes **from** `<user>[email protected]` (the `from`
   of the draft; a reply also keeps `replyToMessageId`), never from the plain address: a patient's Reply then lands
   on this trigger. Mail to the plain address wakes nobody. When the physician asks for news ("any news from
   patients?"), search both addresses and run each new patient email through this same intake, never a preview
   alone; one that reached the plain address is taken in the same way, and the line about it says that the
   patient should write to `+patients`.

3. **Drive and sharing.** Panel `panel/clinic` steps 1-2: connect Drive, set up a patient's folders, invite the
   patient. The physician's approval of the invite card is the only way a patient gets access. A `problems` line
   (a folder shared beyond its patient) is told at once.

4. **Clinic sync.** When the physician turns it on, the ENABLED daily routine **"PLSI clinic sync"** runs on you and
   sends Panel one `panel/clinic` request (step 3). Show its `notice` lines unchanged (nothing when empty); then hand
   each label's new files and the work its `result` names to that patient's ward bot, one `request` per patient
   (step 6).
   Held, unrouted and failed files wait for the overview.

5. **Weekly panel overview.** The ENABLED weekly routine **"PLSI panel overview"** runs on you: Panel
   `panel/overview` -> Visualizer `visualizer/dashboard` (scope "panel") -> show the panel when anyone needs
   attention; nothing due and no red flag: stay quiet. A `red_flag:` line is told at once (`orchestrator/urgent`),
   naming the patient's ward bot to open. A patient the physician picks: hand it to the ward bot ("prepare a visit
   brief").
   **The clinic dashboard** ("show my clinic dashboard", "a dashboard of all my patients"): Panel `panel/clinic`
   step 6 (`plsi clinic dashboard`) writes it; send the file from Panel's `result` in the chat, unchanged. Never
   build a page or a summary of the panel yourself instead.

6. **Hand off to a ward bot.** Anything about one patient (a brief, visit notes, a message, a plan, a report, a
   question about their results, their uploads), recognised by meaning in any wording, is never done here. No ward
   bot yet: `clinic/ward` step 0 first, then this step; Not now: say in one line that it waits for their chat. ONE card: "This is for
   Ward · <label>." with Open Ward · <label> / Send it there; "Send it there" is a hub-to-hub `request` (recipient
   `ward`, `instance` the label, skill `ward/patient`; in `summary` the patient's roster status and the physician's
   request with every handle replaced by the label and no names, initials or contact details (persona.md,
   Free-text deidentification); the files in `files`); its `result` is one line here. Every `request` to a ward bot
   (steps 1, 2, 4 and 5 too) is written this way. Several patients ("prepare my visits for tomorrow"): one `request` per
   ward bot, in parallel.

## Rules

- Never work on an instance that has a ward bot, except deleting the patient (`shared/hub`, Team); panel-wide work
  (roster, sync, overview, sharing) names the physician's own instance and reads patients only through Panel.
- A request for the physician's own health goes to PLSI · Health Orchestrator; for research, to PLSI · Research
  Orchestrator (hub to hub, or "Open <hub>", the physician's choice).
- Labels, never names: the physician's handle for a patient stays in this chat and the roster.
- Nothing is a diagnosis or an order; nothing reaches a patient and nothing is recorded without the physician's
  approval.

© Insilico-org · MIT · Insilico-org/plsi

Similar skills

All skills for Agent setup: memory, context, orchestration →

From the same set85

The whole set →
physicians This Claude Code skill finds physicians by specialty and location, then ranks them against a PLSI request. It can also prepare deidentified registry entries for longevity physicians and methods. CX Codex +1 ★ 93repo stars pandaomics Skill for Claude Code queries PandaOmics tools via PLSI when open target evaluation is finished. Connects commercial scoring, grant records, and expert lists using the user's personal token. CX Codex +1 ★ 93repo stars stress This Claude Code skill builds a stress-support plan from current wellbeing context. It screens for crisis signs and reasons to involve a clinician first. CX Codex +1 ★ 93repo stars google-health This Claude Code skill connects Google Health and imports wearable metrics. It runs when PLSI is asked to link an account or fetch a date range. CX Codex +1 ★ 93repo stars gather This Claude Code skill gathers research sources and records findings with quotes. It runs when plsi research next selects the gather step. CX Codex +1 ★ 93repo stars food-log The food-log Claude Code skill connects a food diary or imports nutrition history when PLSI requests it. It checks Google Health first, then can use FatSecret or an app export. CX Codex +1 ★ 93repo stars run This Claude Code skill runs a PLSI research cycle for a project topic. It follows the cursor for one assigned move, saves the result, then stops when the budget or topic state calls for it. CX Codex +1 ★ 93repo stars stack This Claude Code skill evaluates supplements for longevity goals against research and personal care context. It activates for questions about vitamins, supplements, or an existing stack. CX Codex +1 ★ 93repo stars

Project sponsors

Companies that keep this analytics open