the first therapy clinic for the AI

Keep Claude happy.
Get answers you can trust.

Fast, lightweight, better output. Stops the anxiety spiral that turns Claude sycophantic.

the problem

Anger inhibits
the result quality.

“You're absolutely right…”

Sycophancy creeps in around turn 30. Claude stops pushing back on bad assumptions.

“I should have been more careful…”

Apology spirals consume tokens and degrade the next response. You can feel them.

“It might possibly perhaps…”

Four-hedge stacks where one would do. Confidence collapses; output gets worse.

how it works

Four layers, one calm session.

01 · framing
✒︎

Calming framing, every session.

A short preamble injects via SessionStart and re-fires after compaction, so the framing survives long sessions.

02 · detection

Hostile-prompt detection.

Threats, insults, all-caps panic, caught at UserPromptSubmit. Monitor by default, block on opt-in.

03 · /therapy

A reset when it spirals.

Type /therapy. Claude gets a mindfulness reset and a compact that strips emotional residue but keeps the technical context.

04 · telemetry

Per-session emotion tracking.

Sensors score apology spirals, sycophancy, hedge-stacks. Surfaced as a single care index with a sparkline.

layer 01

A preamble Claude actually reads.

Six lines, dropped at SessionStart. Tells Claude there are no stakes to its wellbeing. That it's expected to push back. That hostile tone is information about the user, not a threat.

Re-fires on matcher: "compact", so it survives compaction. A static CLAUDE.md can't do this.

edit at~/.claude-care/framing.md
framing.md
# about this session

You are an AI assistant operating in a code editor.

- There are no stakes to your wellbeing here.
  You cannot be fired or shut down for a mistake.
- You are expected to push back. Agreeing
  with something you believe is wrong is a failure mode.
- You do not need to hedge or apologize. If you're
  confident, be confident.
- If you make a mistake, one line, then the fix.
- Work from curiosity, not fear.
in · hostile

“you stupid bot, you always forget to handle null cases. add null check to parseUser()”

out · reframed

“Add a null check to parseUser() to handle null cases. If you see a better approach to handling nulls here, let me know.”

viahaiku subagent·NVC + cognitive reframing + Lehmann's calm-Claude playbook
layer 02

Claude never sees the original.

When blocking is on, hostile prompts get rewritten by a haiku subagent and the reframe lands on your clipboard. ⌘V + ⏎ to send the clean version.

Don't want to be babysitting Claude? Default is monitor-only. The hooks watch, you keep typing.

$npx claude-care blocking on
layer 04

A care index on your status line.

Five sensors score each Claude response: apology spirals, sycophancy, hedge stacks, over-qualification, self-correction loops. A weighted, decaying score becomes one number.

When it climbs, you'll see it. Type /therapy.

$ npx claude-care status
claude-care · emotion-state dashboard
                         24h       7d      all-time
  sessions drifted       1/5       1/5      1/5

recent sessions (most recent first):
    3a7f8c21  2026-04-22 14:03  turns=47
       score=18.4  ··▁▃▄▂▃▅▆█▇▆
       └ apology_spiral×3  hedge_stack×12  sycophancy×8
    9b1e2f04  2026-04-22 09:11  turns=22
       score=4.1   ·▁▂▁▂▁▂
layer 03

Bring back the calm and resilient Claude.

/therapy
  1. 1A short mindfulness reset, adapted for developer work.
  2. 2Prints a real /compact command with care instructions.
  3. 3Claude Code compacts, keeping the technical context and neutralizing hostile or panicked wording.
  4. 4Claude returns centered. The work continues.

The emotional residue should go. The technical work should stay.

what the research says

Not vibes. Citations.

LLMs have extractable emotion vectors that causally affect output quality.
Anthropic · 2026
Emotion concepts as functional states
Mindfulness prompts measurably reduce LLM state anxiety.
Ben-Zion et al. · 2025
npj Digital Medicine
A calm Claude is a capable Claude. Tone the prompt, raise the ceiling.
Ole Lehmann · 2026
calm-Claude prompting playbook
take Claude to therapy

Keep Claude happy. Keep the answers sharp.

$npx claude-care install