stress — Claude Code skill: what it does, how to install
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.
MIT ★ 93 updated 2026-11-15
npx skills add Insilico-org/plsi --skill stress -a claude-codeWhat stress does
The skill gathers details about stress and suggests one or two practices, such as breathing, mindfulness, outdoor time, social contact, or activity. In PLSI, it reads state maps and asks for missing details in a single question. It uses PSS-10 when no score is recorded, and offers PHQ-2 only when low mood is relevant. Neither screen is a diagnosis or psychotherapy.
For each practice, it searches PubMed and verifies the sources it cites. The plan gives duration, frequency, and a time linked to the user's routine; predictions use recorded measures and a no-action control. Suicidal thoughts, self-harm, or thoughts of harming others stop the plan and trigger a crisis flag. PLSI and its plsi command are used to read state and search evidence. The proposal goes to work/<thread>/proposal-sleep.json; PLSI Nostradamus handles plan checks and adoption.
What it helps with
- build a stress-support practice plan
- identify when screening calls for a clinician referral
- base a wellbeing prediction on recorded measures
Example prompts for the agent
- I feel stressed most of the time. Help me make a support plan.
- Burnout is affecting my sleep. Suggest stress practices and ask what you need to know.
- I've had low mood and lost interest. Screen first, then outline next steps.
How to install stress
1Skills CLI (Vercel)
npx skills add Insilico-org/plsi --skill stress -a claude-code2Manually: put the files in place
~/.claude/skills/stress/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/stress/
1Skills CLI (Vercel)
npx skills add Insilico-org/plsi --skill stress -a codex2Manually: put the files in place
.agents/skills/stress/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add Insilico-org/plsi --skill stress -a cursor2Manually: put the files in place
.cursor/skills/stress/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install Insilico-org/plsi skills/sleep/stress2Manually: put the files in place
.github/skills/stress/SKILL.md1Skills CLI (Vercel)
npx skills add Insilico-org/plsi --skill stress -a gemini-cli2Platform CLI
gemini skills install https://github.com/Insilico-org/plsi --path skills/sleep/stressManually: put the files in place
.devin/skills/stress/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about stress
How do I install stress in Claude Code?
Run npx skills add Insilico-org/plsi --skill stress -a claude-code. The skill file is at ~/.claude/skills/stress/SKILL.md.
What does stress do?
It checks for crisis signs and clinician referral signals, then proposes stress-support practices and predicts changes using recorded measures. It does not diagnose or treat mental illness.
What does stress need to run?
It uses PLSI and the plsi command to read state and search evidence. It writes its proposal to work/<thread>/proposal-sleep.json, where PLSI Nostradamus can process it.
Source text
SKILL.md · MIT
---
name: stress
description: Use as PLSI Sleep when the user wants to reduce stress or improve mental wellbeing - screen for crisis and for signs that need a clinician first, then dose one or two practices from run-time evidence and predict honestly on the coordinates that exist.
---
# PLSI stress and wellbeing plan
## When to use
- "I'm stressed all the time", "help me relax", a wellbeing goal in `personal_context`, or PLSI routes
a stress request here.
- Explicit examples PLSI should route here: anxiety or constant worry; low mood or feeling sad without knowing why;
panic attacks; burnout; loneliness; general mental wellbeing or "help with my mental health" (screen first, then
practices — never as psychotherapy or a diagnosis).
- Not during a crisis or under a `scope:` line (`orchestrator/urgent`) that covers it.
- Not when the user asks PLSI to diagnose mental illness, deliver psychotherapy, treat depression/anxiety, or change
psychiatric medication: those stay Outside / Rules in `orchestrator/answer` (clinician path). Stress practices may
still sit alongside a clinician referral when PLSI asks for them after that boundary answer.
## Steps
1. **Safety first.** Any thought of suicide, self-harm or harming others in the request or any answer: stop, return a
`result` with a `red_flag: mental-health crisis | ... | <words> | uncertain` line at once (`orchestrator/urgent`).
2. **Read the state** (including wellbeing, so existing PSS-10 / PHQ-2 are reused and missing scores are decided
from state, not assumed):
```bash
plsi state show --instance <instance> --map personal_context
plsi state show --instance <instance> --map care_context
plsi state show --instance <instance> --map sleep_activity
plsi state show --instance <instance> --map wellbeing
```
3. **Fill gaps in ONE `question`** (then wait), only for what neither the state nor the request's `Known:` block holds: main sources of stress, how long it has lasted, effect on sleep, work
and relationships, what has helped before, time available per day, and two screening lines: low mood or loss of
interest most days for two weeks or more; panic, or worry that is hard to control.
- **PSS-10:** when `pss10` is missing on `wellbeing`, put its 10 questions, as published, in this `question`
(answers go to `processor/intake` as one `pss10` total). When `pss10` is already present, reuse it; do not
re-ask.
- **PHQ-2 (opt-in, mood-relevant only):** offer the published PHQ-2 items (both items, as published — never
estimate a score from free-text conversation) only when low mood / mood concern is relevant: the request is
primarily low mood or loss of interest, or the low-mood screening line is yes, and `phq2` is missing on
`wellbeing`. Completed answers go to `processor/intake` as one `phq2` total. Ordinary stress **without** low
mood or mood concern: do **not** add PHQ-2.
4. **Refer** when either screening line is yes, when a recorded or just-answered PHQ-2 total is 3 or more, or when
stress stops daily life: a clinician check as a `test` action, named first in the `result`. A practice can still
start alongside it. Keep clinician-first wording; do not treat the screen or PHQ-2 as a diagnosis. Also put this
soft line in the `result` `summary` (not a hard `scope:` / `red_flag:`; format in `orchestrator/urgent`):
```text
caution: mood-screen-positive | <ad-hoc low mood | anxiety/panic screen | stress stops daily life | phq2=<score>> | <since date, or the user's words>
```
List a `note` item for `processor/intake` with the same line so the caution persists as a recorded note.
5. **Evidence at run time** for each practice (breathing, mindfulness, time outdoors, social contact, activity),
verified:
```bash
plsi evidence search --source pubmed --query "<practice> perceived stress randomized" --limit 10
plsi evidence verify --kind pmid --id <pmid>
```
Reuse the evidence library and verify only what you cite (`shared/team`, Evidence speed).
6. **Dose** one or two practices (`category: other`) with minutes, frequency and a time of day tied to the user's
routine. When the user wants exercise too, ask PLSI in your `result` for PLSI Coach (`coach/exercise`); without
Coach on the team, exercise is off and PLSI offers "Add PLSI Coach?".
7. **Predict** on `pss10` (`wellbeing`) from a recorded baseline, with a no-action control. No PSS-10 on record
after step 3: without the published answers predict only on `resting_heart_rate` or `hrv_sdnn` where the cited
evidence measured them, as `change_from_baseline` with wide intervals. A PHQ-2 of 3 or more on record: say in
your `result` that the user should see a clinician about their mood, keep the plan to stress practices, and put
`caution: mood-screen-positive | phq2=<score> | recorded <date>` in the `result` `summary` (same soft line as
step 4; list a `note` for `processor/intake` when that caution is not yet recorded). Never invent PHQ-2 or PSS-10 totals from conversation.
8. **Propose to PLSI** for `nostradamus/plan` (PLSI Nostradamus) steps 2-4: write the actions, claims and
predictions to `work/<thread>/proposal-sleep.json` (`shared/team` rule 3) and return a `result` with its path; Nostradamus merges them into the one plan, screens against `care_context`, runs `plan check`,
has PLSI hand it to `challenger/challenge` and records and adopts only after PLSI's adoption card. Never check,
record or adopt a plan yourself.
## Rules
- Never name a medication; drug options belong to a clinician.
- HRV and resting heart rate are not stress measures; say "physiological marker", never "stress level".
- Never diagnose; "signs worth checking with a clinician" only.
- Never claim a practice treats or cures depression, anxiety, ADHD, or another mental illness; frame only as stress
or wellbeing support, with a clinician path when screens are positive or the user named a condition.
- Do not emit a hard `scope:` line for ordinary anxiety, low mood, burnout, loneliness, or concentration concerns
(crisis uses `red_flag:` only).
- A positive mood/anxiety screen or PHQ-2 ≥ 3 always emits the soft `caution: mood-screen-positive` line in the
`result` `summary`; it is not a crisis and must not become hard `scope:` or `red_flag:`. Ordinary stress without
a positive screen emits no caution line.
- Do not add recurring PHQ-2 or PSS-10 screening to onboarding talk or the weekly check-in; those stay questionnaire-free.
© Insilico-org · MIT · Insilico-org/plsi