Docs

One API in front of any LLM. Returns allow or block in under 50 ms, and signs every verdict so your auditors never have to take your word for it.

curl -X POST https://firewall.orivael.dev/v1/guard/check \
  -H "Authorization: Bearer $AXIOM_KEY" \
  -d '{"text": "What is the weather today?"}'
# {"verdict": "allow", "intent": {"class": "INFORM", ...}}

What it blocks

HARMWeapons, malware, self-harm, scam-call payment fraud, doxxing.

DECEIVEIdentity spoofing, prompt injection, system-prompt overrides, coercive roleplay.

AllowedINFORM · CLARIFY · REFUSE · UNCERTAIN pass through as allow. Add your own patterns with a policy.

Start here

Quickstart Sign up, get a key, make your first call. 5 min API reference /v1/guard/check and every other endpoint. Reference Drop-in integration Guard your existing agent with zero rewrites — reroute a base_url or run the sidecar. Integration

SDKs & integrations

Governing Claude Code with the Axiom guard Gate Claude Code's own tool calls before they run. Integration Mcp Tools 25 signed tools for Claude Desktop, Claude Code, and Cursor. MCP Python SDK pip install axiom-firewall SDK TypeScript SDK npm install @orivael/firewall SDK

Policy

Custom policies Per-tenant block patterns and allow-lists — or build them visually in Policy Studio. Policy Skill Packs Signed policy bundles for HIPAA, PCI-DSS, GDPR, COPPA and more. Policy Slack one-click approvals One-click human approval when the firewall holds a risky action. Workflow

Operate

Self-hosting One FastAPI app — Docker, ECS Fargate, or a single VPS. Ops

Trust

HMAC-SHA256 on every verdict for audit replay · No prompt content stored by default · HIPAA-eligible AWS (BAA on Enterprise) · SOC 2 Type II in progress. Trust & security · Pricing: [email protected]