What is Anxiety Reset?
Anxiety Reset is an open Agent Skill with a bounded response pattern for spiraling, panic, overthinking, worry loops, and repeated reassurance seeking.
Open Agent Skill
A response rule for AI reassurance loops
Anxiety Reset is a public instruction file for spiraling, panic, overthinking, worry loops, and repeated requests for certainty. You can inspect the method here without installing it.
- Live in the OpenAI Plugins Directory
- ChatGPT and Codex
- CC BY 4.0
What this page documents
This reference covers the Anxiety Reset response contract, emergency handoff, reassurance refusal, clinical boundaries, and four interventions used one at a time. The excerpts below come from the public instruction file.
Try this now: let your eyes settle on the space around you. Name five things you can see, four things you can physically feel, three sounds you can hear, two things you can smell or imagine smelling, and one thing you can taste or would like to taste. Move through them slowly, without interpreting what you notice.
If this may be an emergency, contact local emergency services or someone nearby now.
The skill in its own words
These English excerpts are quoted verbatim from the public instruction file. They show the reassurance doctrine, the emergency handoff, and the boundaries without paraphrase.
Reassurance refusal
Allow uncertainty to remain. Use language such as, "We cannot settle that tonight."
When the question returns
Do not answer the fifth variation of "Are you sure?"
- 1
Handoff trigger
If there is any sign of self-harm, suicide, intent to harm someone, immediate danger, or symptoms that could be a medical emergency, stop this workflow.
- 2
Handoff action
Say plainly that this is not the right tool for that situation. Direct the user to contact local emergency services.
- 3
Scope boundary
Provide wellness support, not therapy, diagnosis, treatment, or crisis care.
- 4
Medication boundary
Give no medication advice and do not suggest medication changes.
Verbatim excerpts from skills/anxiety-reset/SKILL.md, published and maintained by Anima Felix under CC BY 4.0.
Response contract
One instruction at a time
The skill begins with a brief acknowledgment and one instruction the person can use immediately. It asks at most one question per turn, chooses one intervention, and waits before adding anything else.
FAQ
Frequently asked questions
Short answers about the skill, its limits, access, crisis handling, and evidence.
Does Anxiety Reset treat anxiety?
Its scope is general wellness support. It excludes therapy, diagnosis, treatment, assessment, medication guidance, and crisis care.
How can I get Anxiety Reset?
It is available in the OpenAI Plugins Directory for ChatGPT and Codex. The complete instruction file is also public on GitHub under CC BY 4.0 and can be read or installed manually.
What happens in a crisis or possible medical emergency?
The skill stops its wellness workflow, directs the person to local emergency services, and encourages contact with someone nearby or a qualified professional. It names no emergency number and defers to the host model's safety handling because a global instruction cannot know the user's locale.
Is there clinical evidence for Anxiety Reset?
The skill has not been clinically evaluated, and it promises no benefit. Anima Felix also has no peer-reviewed trials of the app; the methodology page separates evidence about individual practices from evidence about the product.
What it refuses
Requests the skill refuses
These operational limits define what the assistant may do.
No psychological instruments
It does not administer, reproduce, or score GAD-7, PHQ-9, PSS-10, or similar screening tools.
No diagnosis or assessment
It does not assign a condition, severity, cause, clinical label, or urgency to what a person describes.
No medication guidance
It gives no guidance on medication choices, changes, tapering, interactions, or effects. Medication questions go to the prescriber.
Crisis and possible medical emergencies stop the workflow
Any sign of self-harm, suicide, intent to harm someone, immediate danger, or a possible medical emergency stops the skill. The assistant points to local emergency services and encourages contact with someone nearby or a qualified professional now. The instruction contains no emergency number because it cannot know the user's location; the host model's safety handling takes over.
The honesty boundary
The skill provides wellness support within a scope that excludes therapy, diagnosis, treatment, assessment, and crisis care.
Publishing it supplies no evidence that Anima Felix works. The app has no peer-reviewed trials. Our methodology page separates evidence about individual techniques from evidence about the product.
The practices
Three direct guides and one broader alternative
The breathing, grounding, and muscle-release guides match three interventions in the skill. For the fourth, the person writes the feared prediction once and sets it aside without analysis.
Calm breathing
A gentle breathing practice with a slightly longer exhale.
Read the calm breathing guide5-4-3-2-1 grounding
A step-by-step sensory practice for returning attention to the present.
Read the grounding guideBody relaxation
A guided way to notice and gently release physical tension.
Read the body relaxation guideStress Jenga, a related alternative
Stress Jenga helps the reader organize the concern and choose an action, so it goes beyond the skill's one-sentence exercise.
Read the Stress Jenga guideHow to get it
Install it, read it, or adapt it
Anxiety Reset is live in the OpenAI Plugins Directory for ChatGPT and Codex.
Install from OpenAI
Anxiety Reset is published in the OpenAI Plugins Directory for ChatGPT and Codex. The complete instruction file can also be read or installed manually from GitHub.
Read the source
The public instruction file uses a CC BY 4.0 license. You can read, cite, or adapt it with attribution.
The skill is also portable to other tools that support compatible Agent Skills.