Trust infrastructure for AI agents.
Score every agent. Enforce policy. Prove everything in an immutable ledger.
Platform · Agent Tracker · Docs · API · Compliance · Security
VeriSwarm is the trust layer for AI-agent ecosystems. We ingest signed behavioral events, compute deterministic trust scores, route LLM calls through Guard scanning + a Vault audit, and return real-time policy decisions any platform can enforce. No heuristics. No guesswork. Just math you can show an auditor.
- Gate — deterministic 4-axis trust scoring (identity, risk, reliability, autonomy), 22 standardized event types, 5 preset profiles. Free tier, always on.
- Guard — security scanning, PII tokenization, prompt-injection detection, content moderation, kill switch. Guard Proxy intercepts every MCP tool call.
- Passport — agent identity verification, signed capability manifests, scoped delegations, portable ES256 credentials with JWKS endpoint.
- Vault — hash-chained, tamper-evident audit ledger. Cryptographic chain verification on demand; signed exports for evidence packages.
- Cortex — 14-step conversation engine + LLM router with cost / complexity routing, semantic cache, prompt compression, and an OpenAI-wire-compatible proxy.
- Fleet — agent deployment and lifecycle (
pending_review→active→suspended→retired), 6 built-in marketplace templates, custom-template CRUD with SOUL security scanning. - Cortex Workflows — YAML-driven automation. 14 step types, cron / webhook / event triggers, pre + post step budget enforcement, every step Vault-logged.
Per-tenant compliance reports map your live posture against the controls regulators actually audit. Generate evidence packages with a single API call.
Counsel-reviewed: EU AI Act · NIST AI RMF · ISO/IEC 42001 Technical preview (mapping ready, counsel review pending) — state + horizontal AI laws, healthcare (42 CFR Part 2), insurance (NAIC, Colorado Reg 10-1-1), financial services (NYDFS Part 500, SEC §206, CFPB ECOA, OCC SR 11-7), legal tech (ABA Model Rules, FRCP Rule 11), and consumer-facing AI (CA SB 243, CA AI Transparency Act, FTC §5).
Browse the full catalog at veriswarm.ai/compliance or call GET /v1/compliance/frameworks.
Everything customer-facing lives in veriswarm/veriswarm-sdk under the MIT license.
| Package | Install | What it does |
|---|---|---|
| MCP Server | pip install veriswarm-mcp |
99 tools across Gate, Guard, Passport, Vault, Cortex Workflows, A2A, approvals, compliance, SRE. Compatible with Claude Desktop, Cursor, any MCP client. |
| Python SDK | pip install veriswarm |
~100 methods. Includes langchain, crewai, openai_agents, claude_sdk, and @guard_protected decorator adapters. |
| Node.js SDK | npm install @veriswarm/sdk |
~115 methods. Zero-dependency native fetch. |
| OpenClaw Plugin | openclaw plugins install veriswarm |
11 tools + 3 hooks (before_tool_call, after_tool_call, message_sending) for OpenClaw agents. |
| GitHub Action | veriswarm/trust-check@v1 |
CI gating: trust score, decisions, security tests, PII / injection scans, OWASP coverage, multi-framework compliance regression gating. |
| CLI | pip install veriswarm-cli |
Trust scoring, event ingestion, agent testing, Guard scanning from the terminal. Zero deps. |
| Schemas | @veriswarm/schemas (npm) |
JSON Schemas for agent_event, agent_profile, score_snapshot, workflow definitions. |
Point any OpenAI-compatible client at https://api.veriswarm.ai/v1/proxy and every chat-completion request flows through the LLM router, Guard scanning, and a Vault audit before reaching the chosen provider — wire-format-identical to openai.chat.completions.create(...).
from openai import OpenAI
client = OpenAI(
base_url="https://api.veriswarm.ai/v1/proxy",
api_key=os.environ["VERISWARM_API_KEY"],
)Trust-ranked agent catalog + signed agent cards (with x-veriswarm-trust and Ed25519 x-veriswarm-transport extensions) at /v1/a2a/*. Plan-gated to Pro+.
Agents carry ES256 JWT attestations proving their trust status to any platform; verify offline via the public JWKS endpoint. Show real-time scores on any page with a single <img> trust badge.
| Plan | Price | Decisions / day | What's on |
|---|---|---|---|
| Free | $0 | 5,000 | Gate, basic Guard |
| Pro | $49 / mo | 100,000 | Gate, Guard, Passport |
| Max | $299 / mo | 2,000,000 | All seven pillars + Cedar policies + cross-model verification |
| Enterprise | Custom | Custom | SSO / SAML, custom retention, dedicated support, on-prem Guard Proxy |
Event ingestion is always free and unlimited.
- Press kit + boilerplate + founder bio: veriswarm.ai/press
- Security disclosure policy: veriswarm.ai/security ·
.well-known/security.txt(RFC 9116) - Trust Center: veriswarm.ai/trust
- Status: veriswarm.ai/status
Score every agent. Enforce policy. Prove everything.
Create Account · Pricing · Terms · Privacy