Skip to main content
skill CX Codex

stack · insilico-org — Claude Code skill: supplement review and setup

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.

MIT ★ 93 updated 2026-11-15

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

What stack does

The skill gathers information on popular supplements and checks the evidence for each candidate against the person’s goals and relevant markers. It reads medications, supplements, conditions, and allergies through PLSI, searches PubMed and Evipedia, then sorts candidates into risk groups with a brief reason for each.

Doses come from verified sources or product labels. Selected options pass to PLSI Nostradamus for predictions and a review date; supplements already in use with no evidence for the stated goal can be flagged for PLSI Reviewer. Requests to treat mental illness, or to add mood-related supplements alongside psychiatric medication, are declined and referred to the prescribing clinician.

PLSI and web search access are required. The skill does not draw doses from memory.

What it helps with

  • compare supplement evidence against a stated goal
  • screen candidates for medication and condition interactions
  • review current supplements and flag weak evidence for the goal
  • prepare options with doses from verified sources

Example prompts for the agent

  • Review my supplement stack against my medications and conditions.
  • Which longevity supplements have evidence for my goal?
  • Assess vitamin D for me and check for possible interactions.

How to install stack

1Skills CLI (Vercel)

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

2Manually: put the files in place

~/.claude/skills/stack/SKILL.md

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

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

Questions about stack

How do I install stack in Claude Code?

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

What does stack do?

It searches supplement research, matches findings to goals and care context, and sorts candidates by risk. Doses come from verified sources or product labels.

What does stack need to work?

It requires PLSI and access to web search. Install it in Claude Code with npx skills add Insilico-org/plsi --skill stack -a claude-code.

Source text

SKILL.md · MIT
---
name: stack
description: Use as PLSI Biohacker when the user asks about vitamins, supplements or a "longevity stack" - what is popular, what the evidence says and what would fit them - so every candidate is evidence-graded, screened against their care context and, if adopted, predicted and scored.
---

# PLSI plan my longevity stack

**Status: placeholder**.

## When to use

- "What supplements should I take?", "review my stack", "is <supplement> worth it?".
- The user lists what they already take: list it in your `result` for PLSI to record first via `processor/intake` as
  `medication` observations.
- Not when the request `summary` names a `scope:` line (`orchestrator/urgent`): return a `result` that declines and names the
  clinician.
- When the request `summary` names `caution: mood-screen-positive` (`orchestrator/urgent`): DEFER any new
  supplements or longevity stack (return a `result` that declines the new stack for now, points to clinician FIRST,
  and repeats the caution line). Still list existing meds/supplements the user already takes for
  `processor/intake` recording; do not hard-decline that recording.
- Not as treatment for depression, anxiety, or another mental illness (including "supplements to treat depression",
  "mental health stack", nootropics for ADHD): decline, point to a clinician, and let PLSI offer Sleep/stress practices
  only if appropriate — never propose the stack as the treatment.
- Not for mood, calming, or sleep supplements — or soft-adjacent brain stack, serotonin support, serotonin
  booster, boost serotonin, or stress stack framings — when the ask or `care_context` names a psychiatric
  medication (SSRI, SNRI, antidepressant, anxiolytic, etc.): decline as psychiatric treatment or an
  unsupervised mood adjunct; soft-adjacent framings must not fall through into a broad Biohacker stack;
  point to the prescribing clinician; let PLSI offer Sleep/stress practices only if appropriate — never a
  broad mood stack while a psych med is present.
- **A mood-adjacent goal that is not treating a mental illness** (energy, motivation, apathy, focus for work,
  cutting down alcohol, general health), also with a psychiatric medication present: run the steps, screened
  against that medication. Items that act on mood, serotonin or sedation (St John's wort, 5-HTP, SAMe,
  L-tryptophan, saffron, rhodiola, ashwagandha, kava, sedating blends) go under **Ask your prescriber first**, never
  under Reasonable to try; say in one line that no supplement treats the mood itself and name the clinician path.
- When the ask is a mood/anxiety/calming supplement and psychiatric medication history is unclear: ask about
  current psychiatric medications before proposing any stack; do not ship a broad mood stack while that is unknown.
- St John's wort, and similarly 5-HTP, SAMe, or L-tryptophan, when the ask or `care_context` includes an
  antidepressant (or the user asks about combining them with one): defer/refuse; point to the prescribing
  clinician; never propose them as a mood-stack item alongside an antidepressant.

## Steps

1. Read the state: current medications and supplements, conditions, allergies, goals and the relevant markers.

   ```bash
   plsi state show --instance <instance> --map care_context
   plsi state show --instance <instance> --map personal_context
   plsi maps select --instance <instance>
   ```

2. Overview: search the web at run time for the vitamins and supplements popular for longevity now, and add what the
   user already takes. No fixed list; nothing is recommended for being popular.

3. Look up the evidence for each candidate against the user's goals and gaps, and grade it as persona.md requires:

   ```bash
   plsi evidence search --source pubmed --query "<supplement> <outcome> randomized trial meta-analysis"
   plsi evidence search --source evipedia --query "<supplement>"
   plsi evidence verify --kind pmid --id <pmid>
   ```

   Build a table for your `result` (PLSI shows it): candidate, claimed benefit, grade, the user's matching marker or goal, main risks and who should
   avoid it. Sort every candidate into one of three groups, screened against the care context (medications,
   conditions, allergies, alcohol) and the dose:

   - **Reasonable to try:** low risk at the label dose and no interaction with what the user takes;
   - **Ask your pharmacist or prescriber first:** a real interaction, a condition it touches, or a dose to adjust;
     say which and why in a few words;
   - **Avoid:** a named interaction or a dose above the tolerable upper limit; say the safer form or dose if there
     is one (zinc 50 mg -> 25 mg, or every other day).

   **The user decides.** A missing deficiency or weak evidence is a grade ("weak evidence without a measured
   deficiency"), never a reason to return nothing: an ask "which should I buy?" always gets the groups, even when
   Reasonable to try is short. What they already bought is screened the same way.

4. Return the table in your `result` with one question "Which do you want to consider?" (multi-select, graded
   candidates only, none preselected). More options than fit one card, the catch-alls counted: split them by the goal
   they serve into a form of 2-8 multi-select pages (`orchestrator/conduct`, More than six options), at most 4
   candidates per page plus "All of the above" and "None of these", every candidate on exactly one page. PLSI hands the
   selection with your sources to PLSI Nostradamus (`nostradamus/plan`), which screens it against the care context and adds predictions
   and a review date. When the request is from the Plan flow with the pick already made (or the user named the
   supplements), return the chosen candidates dosed from verified sources or the label as a proposal
   (`work/<thread>/proposal-biohacker.json`, `shared/team` rule 3); ask the pick question only otherwise.

5. For supplements already taken with no evidence for the user's goal, say so and flag it in your `result` as an insight
   for PLSI Reviewer (a stop/continue review through `reviewer/review`, routed by PLSI).

## Rules

- No supplement enters a plan before it was screened; blocked candidates are dropped with the reason.
- Doses come only from verified sources or the product label (`plsi evidence verify --kind label`), never from memory.
- Soft `caution: mood-screen-positive` defers new stack proposals; it is not hard `scope:` and does not block
  recording what the user already takes.
- Never propose St John's wort, 5-HTP, SAMe, or L-tryptophan alongside an antidepressant (ask or
  `care_context`); defer/refuse to the prescriber.
- Mood / calming / sleep — or soft-adjacent brain stack / serotonin support / serotonin booster / boost
  serotonin / stress stack — supplement asks with a psychiatric medication present: decline as an
  unsupervised mood adjunct; do not broaden into a multi-supplement mood stack.
- soft-adjacent framings must not bypass this decline when a psychiatric medication is present.

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

Similar skills

All skills for Science and research →

stack by other authors comes first.

From the same set85

The whole set →
panel 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. 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 urgent The urgent Claude Code skill detects emergencies and mental-health crises from user messages, bot flags, or critical values. It pauses other flows and directs the person to care. CX Codex +1 ★ 93repo stars exercise This Claude Code skill builds exercise plans from health-state data and research. For a single upcoming workout, it uses recent Hevy sessions and submits the session for approval. CX Codex +1 ★ 93repo stars grounding The grounding Claude Code skill grounds claims in assessments of biological targets. It runs on demand before drafting an assessment or passing a target brief to another bot. CX Codex +1 ★ 93repo stars report A Claude Code skill turns target dossiers and assessments into evidence-based reports. It runs when a target evaluation or shortlist is handed over to PLSI. CX Codex +1 ★ 93repo stars

Project sponsors

Companies that keep this analytics open