May 9
Loading video player...
SupraWall is the open-source security perimeter for AI agents. Every tool call your agent makes — every API hit, every shell command, every database query — is intercepted before execution, evaluated against a deterministic policy, and either allowed, denied, or sent for human approval. Every decision is recorded in an RSA-signed audit trail your auditors and regulators can verify. If Paperclip gives your agents a company, SupraWall gives them a constitution. — — — INSTALL pip install suprawall-sdk npm install @suprawall/sdk THREE-LINE QUICKSTART (Python) from suprawall import secure_agent from my_app import build_agent agent = secure_agent(build_agent(), template="hr-employment") LINKS GitHub → https://github.com/wiserautomation/SupraWall Website → https://suprawall.com Docs → https://suprawall.com/docs Discord → https://suprawall.com/discord License → Apache 2.0 — — — WHAT SUPRAWALL IS - A deterministic security perimeter for AI agents - Framework-agnostic: wraps Paperclip, LangChain, CrewAI, AutoGen, Claude Code, OpenClaw (via MCP), and the Vercel AI SDK - Self-host the open source, or use the managed cloud - Ships with pre-built sector templates for the 8 EU AI Act Annex III high-risk categories plus DORA for finance WHAT SUPRAWALL IS NOT - Not an agent framework. We protect agents; we don't orchestrate them. - Not an LLM guardrail. We enforce policy on actions, not on prompts. - Not an observability tool. Datadog watches. SupraWall stops. - Not AI-powered. The enforcement layer is deterministic by design. EU AI ACT — 9 SECTOR TEMPLATES SupraWall ships pre-built policy templates mapped to the EU AI Act Annex III high-risk categories (enforcement deadline: August 2, 2026): 1. Biometrics Annex III(1) 2. Critical Infrastructure Annex III(2) 3. Education Annex III(3) 4. HR & Employment Annex III(4) 5. Healthcare Annex III(5) 6. Law Enforcement Annex III(6) 7. Migration & Border Annex III(7) 8. Justice & Democracy Annex III(8) 9. Banking & Finance (DORA) Each template encodes Article 9 (risk management), 13 (transparency), and 14 (human oversight) requirements as deterministic policy rules. Swap templates with a single argument: secure_agent(agent, template="healthcare"). CHAPTERS 0:00 The problem 0:08 Agent frameworks are permissive 0:16 SupraWall intercepts the call 0:24 Any agent stack, inside the perimeter 0:32 EU AI Act — the 9 sector templates 0:40 Install it — — — ABOUT SupraWall is built by Wiser Automation. Founder: Alejandro Peghin. Open source under Apache 2.0. Self-host forever, or let us run it. If you ship AI agents in production — or are about to — star the repo. When an autonomous agent eventually does something dumb, deterministic enforcement is the difference between a bad day and a board-level incident. https://github.com/wiserautomation/SupraWall — — — #AIAgents #OpenSource #AgentSecurity #EUAIAct #DORA #AISafety #LLMOps #LangChain #CrewAI #AutoGen #Paperclip #AIAgentSecurity #Cybersecurity #DevTools #InfoSec