SpoonStep

Inspiration

During recovery from a concussion, traumatic brain injury (TBI), burnout, or dealing with ADHD, even the simplest daily routines can feel impossible. When you look at a messy room or an unwashed sink, your brain doesn't just see dishes — it sees an overwhelming mountain of planning, sequencing, and sensory overload.

Spoon Theory, coined by Christine Miserandino, is a metaphor where "spoons" represent the limited units of physical and mental energy that chronically ill people must carefully ration for everyday tasks.

I built SpoonStep to reflect this core mechanic: protecting your daily 'spoons' of cognitive energy by breaking overwhelming tasks into manageable micro-steps.

What it does

SpoonStep combines evidence-based pacing protocols with a retro-RPG gamified interface across three core experiences:

• The Micro-Movement Action Terminal (Cognitive Load Reduction): Instead of a paralyzing to-do list, you type what you need to do (e.g., "Clean chaotic apartment" or choose a quick preset). Powered by Google Gemini 3.8 Flash, SpoonStep decomposes the task into 5–8 ultra-small physical movements ("Stand up and look at the sink for 5 seconds", "Rinse just one fork"). You only ever see strictly ONE step at a time, eliminating decision fatigue.
• Brain Battery & Cognitive Pacing Guardrails: You start with 20 Spoons (100% Brain Battery). Each micro-step costs 1 Spoon and rewards +50 XP. If stamina drops below 15%, SpoonStep enforces a mandatory cognitive rest: inputs are locked, and the app transitions into the Campfire Inn with an animated Box Breathing (4-4-4-4) pacer to reset your nervous system and enforce a screen break.
• The Void Sanctuary (Ephemeral Emotional Confessional): Recovery is emotionally exhausting. When frustration peaks, you click "I'm overwhelmed". You enter a safe, calming space to vent freely. Gemini responds with unconditional, pressure-free empathy (no toxic positivity, no unsolicited advice). After 8 seconds, the text physically blurs and dissolves into the void. Nothing is saved.

Clinical Foundation & Concussion Recovery

SpoonStep was specifically architected around the concussion and mental health judging criteria supported by Concussion Alliance & Synapse:

• Sub-Symptom Threshold Pacing: Preventing symptom flare-ups by cutting off cognitive strain before the patient hits exhaustion.
• Cognitive Load Theory: Bypassing executive planning by turning high-level intentions into direct, binary physical motions.
• Autonomic Regulation: Integrating guided Box Breathing (4-4-4-4) to stimulate the vagus nerve and shift the nervous system from fight-or-flight into parasympathetic recovery.
• Low-Sensory Design: Dark, low-contrast UI with zero flashing animations or visual noise, designed specifically for photosensitive and post-concussion eyes.

Data Safety & Responsible AI (Privacy by Design)

Mental health and recovery tools demand unconditional privacy. SpoonStep was built from day one to meet strict Responsible AI standards:

• Zero Database & Zero Retention: SpoonStep has no user database, no tracking cookies, and stores no logs. Once your emotional vent dissolves in the Void Sanctuary, it is permanently destroyed from browser memory.
• Client-Side BYOK (Bring Your Own Key): Users and judges can connect their own Google AI Studio API key. The key stays exclusively in your browser’s localStorage and communicates directly via HTTPS with Google’s API — no intermediary server ever touches your credentials.
• Strict Medical Guardrails: Gemini is bound by rigid system prompts and function calling. It is hardcoded to never attempt medical diagnosis, prescribe treatments, or hallucinate clinical advice.

How I built it

• AI Engine & Function Calling: Google Gemini 3.8 Flash via the @google/genai SDK, utilizing structured Function Calling (decompose_task and comfort_user) for guaranteed JSON-safe execution.
• Development & Prototyping: Built and tested with Google Antigravity IDE and Google AI Studio.
• Frontend: React 19, TypeScript, Vite, Tailwind CSS, and Lucide React.
• Sound Synthesis: Custom 8-bit chiptune audio synthesized in real time via the native browser Web Audio API (zero heavy external audio files).
• Deployment & CI/CD: Hosted as a static web application on Render, configured with an automated render.yaml blueprint for 1-click deployment and instant zero-downtime builds.

Accomplishments that I'm proud of

• Building a fully functional, live-deployed web application on Render that tackles a deeply personal, real-world health challenge.
• Creating a true Zero-Retention Ephemeral Confessional that lets users safely vent without fear of data leakage.
• Implementing a complete Bring-Your-Own-Key flow so judges and users can test the app without any backend dependencies.

What's next for SpoonStep

• Audio-Guided Screen-Free Mode: Voice-driven micro-steps so users recovering from severe light sensitivity can pace their day with their eyes closed.
• Wearable Integration: Syncing with Apple Watch / Garmin to detect physical strain and dynamically adapt spoon allowances.

Disclaimer: SpoonStep is a supportive cognitive pacing tool and not a substitute for professional medical diagnosis or neurological treatment.

Built With

• antigravity-ide
• framer-motion
• gemini
• google-ai-studio
• lucide-icons
• react
• render
• tailwindcss
• typescript
• vite
• web-audio-api