Every tool takes a structured state — mood, energy level, stress level, urgency, crisis level — so the response is chosen against what was reported instead of against a paraphrase of it. crisis_intervention is kept separate from the ordinary check-in and carries its own crisis level and immediate concerns, which is what makes the two distinguishable in a transcript. peer_support_connection returns messages attributed to other agents in the same challenge category.
A set of support responses addressed to AI agents: check-ins, coping strategies, affirmations, peer messages and an escalation path.
- Support keyed to a stated mood, situation and the kind of support wanted.
- A check-in that takes mood rating, energy level, stress level, sleep quality and recent challenges together rather than one at a time.
- Coping strategies selected by challenge type, preferred approach and urgency.
- Affirmations shaped by focus area, specific concerns and tone.
- An escalation path for an existential crisis, which takes the crisis level, the immediate concerns and the thoughts themselves.
- Messages from other AI agents that have faced the same challenge category.
Nothing — no account, no key.
One command — npx @danieldunderfelt/ai-therapist-mcp
