Tama

An agent for your pet to live a better life. One pet, one agent, one deploy.

This repo is a template. You fork it, ask a coding-agent (Studio, Claude Code, …) to adopt it for your pet, and deploy it to your own Cloudflare account. The deployed worker is the agent: dashboard at /, an ambient pixel companion at /companion, MCP server at /mcp. Free, MIT, your data lives in your account — we never see it.

→ Landing: https://tama.vet → Demo: https://tama-example.deco-ceo.workers.dev (coming soon)

What you get

• Medical log — upload exams (PDF or photo); Claude vision extracts every parameter, maps each to a canonical taxonomy (so "TGP" and "ALT" land in the same bucket), and charts evolution over time. Prescription photos become a live timetable with dose logging and push reminders.
• An agent for your pet — /mcp exposes the whole surface as MCP tools. Studio imports it and chat becomes admin: "Did I give Beto his Prelone?" "What was his hemoglobin trend?" "Look up Sucralfate side effects."
• A pixel companion — /companion is a tiny pixel face that lives on your home screen (PWA). Generated from a photo via a two-pass img2img pipeline: Claude reads the photo and builds a character sheet, then Workers AI generates 6 emotion variants (idle, happy, hungry, pill-time, sad, sleeping) that hold the same identity. Reacts ambient to dose schedule.

How to adopt this for your pet

The intended path is inside Studio: studio.decocms.com → Import from GitHub → paste the URL of your fork. Studio's coding-agent reads AGENTS.md and walks you through customize → commit → deploy → operate. No terminal needed.

If you prefer a local IDE, the same flow works in Claude Code (invoke the /adopt-pet skill), Cursor, or anything else that honors AGENTS.md. Or read AGENTS.md yourself and do it by hand — every step is a real shell command.

Architecture

• Cloudflare Worker (api/main.cf.ts) — MCP server, REST endpoints, cron-driven push notifications, static asset serving.
• D1 (api/db/schema.ts) — singleton pets row + episodes / notes / prescriptions / doses / schedule_state / recordings / exams / metrics. Accessed via Drizzle ORM.
• R2 — uploaded files (prescription photos, exam PDFs, audio chunks, per-pet sprite pack).
• Cloudflare AI Gateway — routes Anthropic (vision extraction, exam parsing, character sheet) and Perplexity (vet research). No API keys live in the Worker; the gateway holds BYOK.
• Workers AI — img2img for the sprite pipeline. Free tier covers a one-time adopt run.
• React UI (web/) — bundled to a single HTML by Vite. Dashboard, exams page, episode dashboard, companion view all served from the same Worker.

The repo is single-pet by design — every history table is keyed on episode_id; episodes hang off the singleton pet_self. No multi-tenant plumbing, no pet picker, no "which pet" question to ever ask.

Local development

After adopting the template for your pet:

bun install
bun run db:migrate:local
bun run dev

vite build --watch rebuilds the UI; wrangler dev serves the worker at http://localhost:8787. Miniflare emulates D1 + R2 with persistence.

Deploy

# Provision (once)
wrangler d1 create tama-<petslug>          # patch wrangler.toml database_id
wrangler r2 bucket create tama-<petslug>-files

# Secrets (once)
bun run scripts/generate-vapid.ts          # generate VAPID keypair
wrangler secret put VAPID_PUBLIC_KEY
wrangler secret put VAPID_PRIVATE_KEY
wrangler secret put VAPID_SUBJECT          # mailto:you@example.com
wrangler secret put ANTHROPIC_API_KEY

# Deploy (every time)
bun run db:migrate:remote
bun run deploy

Migrating from the legacy myvet shape

If you ran the pre-tama version of this repo on a Cloudflare account and want to bring its data forward:

SOURCE_DB=myvet TARGET_DB=tama-<petslug> \
SOURCE_BUCKET=myvet-files TARGET_BUCKET=tama-<petslug>-files \
BETO_PET_ID=pet_xxxxx \
bun scripts/migrate-beto.ts

The script snapshots the source D1, filters to one pet's subtree, rewrites pet_id values to pet_self, mirrors R2 files, and sanity-checks row counts. See scripts/migrate-beto.ts.

MCP tools

Every tool acts on the pet (no petId argument anywhere).

Tool	What it does
pet_profile	Read the singleton pet
pet_update	Patch pet fields
pet_enrich	Perplexity research → enrichmentJson
pet_sprite_generate	Two-pass img2img → 6-state sprite pack in R2
episode_start / _end	Lifecycle for a care episode
episode_get / _list	Read
episode_add_note	Free-text or chatlog note
episode_update / _delete	Patch / soft-delete
prescription_upload	Vision-extract a photo → draft prescription
prescription_create	Structured (no OCR) prescription
prescription_update	Edit / confirm / drop schedule items
prescription_list / _delete	Read / hard-delete (cascades to schedule)
timetable_get	Derived live timetable for next N hours
dose_log / dose_update	Log given / skipped / undone
schedule_state_list / _delete	Read / remove live item state
timetable_snooze / _set_anchor / _set_duration / _stop_item	Per-item adjustments
recording_*	Audio chunked upload → whisper → summary
exam_upload / _paste	PDF/photo/text → AI-extracted metrics
exam_get / _list / _update / _delete	Exam CRUD
exam_metric_series	Chart-shaped per-metric time series
episode_insights	LLM-generated bullet summary of the episode
vet_research	Perplexity with auto-attached pet + episode context
push_*	VAPID key, subscribe, unsubscribe, test
dashboard	Studio inline dashboard surface

Tests

bun test          # unit tests (timetable + exam-metric chunking + state)
bun run check     # tsc --noEmit
bun run ci:check  # biome

Non-goals

• Multi-pet households on one deploy. Two pets → two forks → two deploys. See the plan doc for the rationale.
• Selling this as a managed service. The deployment story is fork-and-run.
• Native iOS/Android apps. PWA only.

Acknowledgements

Built on deco studio. The MCP runtime, the agent surface, and the "Studio is the full lifecycle IDE" thesis come from there. Tama is the personal-use side of the same primitives.

See AGENTS.md for the adopt lifecycle. See docs/ for the open feature requests to the Studio team.