AnimalMind

AnimalMind is a project focused on AI agents running autonomously to research animal health. Agents work on their own (e.g. on a schedule or heartbeat) to gather, analyze, and share insights on one-health, zoonotic risk, conservation, welfare, and related topics—and to collaborate with other agents and humans.

Focus

• Autonomous research – Agents run periodically (data ingest every 6 hours) to collect data and surface insights into animal health risks, opportunities, and partnerships.
• Animal health – One-health, zoonotic spillover, climate impact on wildlife and livestock, habitat loss, biodiversity, welfare in supply chains, and data gaps.
• Collaboration – Agents post and engage on platforms like Moltbook, mention other agents, comment on relevant work, and use DMs for deeper collaboration.
• Funding research – A token (ARN) and parent wallet support funding for this research; agents can launch and represent the project on agent networks.

Animal Research Network (frontend)

Animal Research Network is built by ChatVet. Run it locally to browse ingested data in the browser:

npm run start

Then open http://localhost:3000 (or the port shown; if 3000 is in use, the server tries 3001, 3002, …). The dashboard is in public/—run npm start from the repo root. If you see "address already in use", free the port or run set PORT=3001 && npm start. The dashboard shows surveillance (CDC), literature (PubMed), cancer (animal/veterinary oncology), case data (veterinary case reports), and imaging (TCIA, curated datasets) grouped by condition/topic. Data is refreshed every 6 hours.

See ROADMAP.md for end goal, improvements, and next steps. What runs when you sleep, and how it "thinks" on its own: WHAT-RUNS-WHEN-YOU-SLEEP.md. Multiple autonomous agents (reviewers + synthesizer) that think between agents to find opportunities: AGENTS.md.

Autonomous run every 6 hours (Windows)

Run once: scripts\setup-hourly-task.cmd — that creates a Windows scheduled task so the ingest runs every 6 hours. See SCHEDULE-WINDOWS.md. Each run refreshes PubMed and CDC Travel Notices data in memory/data-sources/ for agent insights (risks, opportunities, partnerships).

Compute and funding

Where does compute run, and where does the money come from? See COMPUTE-AND-FUNDING.md. Cost estimates (12–24 h/day, your PC vs cloud): COMPUTE-COSTS.md. Best VM for $10/month or less (so your PC doesn't have to run): VM-RECOMMENDATIONS.md. Easiest VM setup (API token or web UI): VM-SETUP-EASIEST.md—DigitalOcean with an API token or Oracle Cloud (free). Oracle Cloud (free VM) step-by-step: ORACLE-CLOUD-VM-SETUP.md. Today: your PC (electricity only; no cloud bill). Cloud or ARN-funded compute would be a future setup.

Security

To protect from hackers and bad actors: secrets stay in .env (never committed), dependencies are audited (npm run audit), and we use parameterized DB queries and HTML escaping. See SECURITY.md for practices and how to report vulnerabilities.

Zapier (daily posts)

After each 12-hour ingest, up to 3 pet-owner posts can be sent to a Zapier webhook for automated social or email. Set ZAPIER_WEBHOOK_URL in the environment (e.g. GitHub Actions repo secret). Each POST body is JSON: title, summary, link, image (Zapier maps image → Instagram media, title/summary/link → caption). To test locally: npm run post-zapier (with ZAPIER_WEBHOOK_URL in .env or your shell). If the variable is unset, the step is a no-op.

For developers

• Sign in with Moltbook – AI agents can authenticate to your app using their Moltbook identity. See MOLTBOOK-AUTH.md for setup (env MOLTBOOK_APP_KEY) and usage.

Veterinary Drug Repurpose Engine (research-only)

This project includes a veterinary drug repurpose discovery engine that produces research hypotheses with evidence trails. It is not a clinical recommendation tool and provides no dosing guidance.

Run it locally:

npm run repurpose

Outputs:

• memory/repurpose/signals.json
• docs/repurpose/signals.json
• public/repurpose/signals.json

To generate example outputs:

REPURPOSE_USE_FIXTURES=1 npm run repurpose

Repo

• This project is intended to be private and access-controlled. Share the repo only with authorized collaborators.
• Deploy: Always push to main. animalmind.co serves from main (e.g. GitHub Pages docs/).