Skip to content
VeriSwarm
关于我们
文档定价智能体技能
登录注册
  1. 首页
  2. /Changelog
VeriSwarm
  • English
  • Español
  • Deutsch
  • Français
  • Italiano
  • Português
  • 日本語
  • 한국어
  • ✓ 简体中文

产品

  • 定价
  • 文档
  • API
  • 智能体技能
  • OATS 规范

信任

  • 信任中心
  • 安全
  • 合规
  • 状态
  • 更新日志

公司

  • 关于我们
  • 博客
  • 开源
  • 投资者
  • 新闻

法律

  • 条款
  • 隐私
  • SLA
  • DPA
  • 无障碍
Changelog

What's New

Platform releases, security updates, and feature announcements in reverse chronological order.

2026-04-01

Cortex Workflows, Portable Credentials & PLG Onboarding

New Features

  • Cortex Workflows (Preview) — YAML-driven automation with 14 step types, cron/webhook/event triggers, and budget enforcement
  • Portable credentials on Free plan — ES256 signed JWTs with trust scores, verifiable via JWKS endpoint
  • Tenant-facing Passport — delegations and manifests manageable from account and agent pages (no longer admin-only)
  • Agent-scoped Vault view — inline ledger on agent detail page with event type, source, actor, and hash chain
  • Agent Marketplace docs — template discovery, deploy wizard, custom templates with SOUL security scanning
  • Guard Hooks (Preview) — PII protection for AI coding agents (Claude Code, Gemini CLI, Codex CLI)

Improvements

  • Auto-verify agents on ownership claim — claiming with a valid API key now proves identity
  • Helpful score endpoint responses — 'no events yet' and 'score pending' guidance instead of bare 404s
  • PII detection accuracy — per-type confidence thresholds reduce false positives on URLs and code
  • SDK User-Agent headers — fixes Cloudflare blocking on Python and Node SDKs
  • Pricing aligned across all surfaces — homepage, pricing page, billing settings, and docs now match
  • Mobile responsiveness — scrollable tab bars, wrapping button rows
  • 65+ MCP server tools (up from 39)

Documentation

  • 3 new docs: Cortex Workflows, Agent Marketplace, Guard Hooks (all marked Preview)
  • Compliance language updated — 'certifications in progress' across all references
  • Internal implementation details stripped from customer-facing docs
  • Quickstart enhanced with plan limits, scoring latency guidance, and slug field
2026-03-29

Security Hardening & Fleet Management

New Features

  • Two-tier widget security — Public and secure chat endpoints with widget tokens and JWT sessions
  • Fleet agent management — Deploy, manage, and monitor agents with full lifecycle controls
  • Cortex runtime — LLM routing with Azure OpenAI, RAG knowledge retrieval, and conversation logging
  • Guard UI — Kill switch controls, policy creation, and on-demand security scanning
  • 26 integration adapters — OAuth and API-key based integrations across CRM, email, calendar, and more

Security

  • Comprehensive platform security audit and hardening
  • Per-token-type cryptographic key isolation
  • Encryption at rest for all sensitive data with key stretching
  • PII tokenization on all LLM pathways
  • GDPR right-to-erasure with cryptographic proof
  • Widget origin allowlisting and rate limiting

Platform

  • Zero-downtime rolling deployment architecture
  • Encrypted off-site backups with automated rotation
  • Platform-hosted LLM inference for trials and demos
  • Full vault hash-chain verification
2026-03-25

Initial Platform Launch

Features

  • Gate trust scoring engine with 22 event types and 5 scoring profiles
  • Guard security scanning with injection detection
  • Passport identity verification and signed manifests
  • Vault immutable audit ledger with hash-chain integrity
  • Agent registration, claiming, and ownership transfer
  • Provider billing with Stripe integration
  • SAML SSO and SCIM provisioning
  • MCP server for AI-native integration