Skip to content

Releases: kindrat86/agentmail

Release list

v0.5.0: agentmail Sanctions Exposure Index (SEI)

v0.5.0: agentmail Sanctions Exposure Index (SEI) Latest
Latest

Choose a tag to compare

View all tags
@kindrat86 kindrat86 released this 18 Jul 15:23
v0.5.0
c634487

Changelog

All notable changes to sanctions-mcp (agentmail) are documented here.
The format follows Keep a Changelog, and this project adheres to Semantic Versioning.

[0.5.0] — 2026-07-18

Added

  • agentmail Sanctions Exposure Index (SEI) — proprietary 5-factor framework for quantifying an AI agent's OFAC sanctions exposure. Velocity (30%), Jurisdiction overlap (25%), Asset class (20%), Screening posture (15%), Disclosure readiness (10%). Published in the 2026 Agent-Payment Sanctions Exposure Report.
  • Interactive SEI Calculator at https://sanctionsai.dev/tools/sei-calculator — computes SEI scores in real-time.
  • 36 hallucinated-URL 301 redirects (AI-guessed slugs now resolve to real pages).
  • 8 section-index pages for AEO crawlability (/for, /compare, /integrations, /glossary, /tools, /vs, /how-to, /cost).
  • PostHog AI-referral auto-capture — fires ai_referral_detected events when visitors arrive from ChatGPT/Perplexity/Gemini/Claude/Copilot/DeepSeek/Grok + 10 more AI sources.
  • Self-reported attribution survey on /start squeeze page (14 options including 6 AI sources).
  • Dataset + Article schema for the SEI research report.

Changed

  • knowledge-graph.json SoftwareApplication.alternateName expanded with SanctionsAI, sanctions-mcp, agentmail API for entity consolidation.
  • pyproject.toml keywords now include sei, sanctions-exposure-index, x402.
  • Homepage now at sanctionsai.dev (was github.com/kindrat86/agentmail).
  • Documentation and Changelog URLs added to PyPI metadata.

Fixed

  • robots.txt no longer advertises 404 section-index paths.
  • Sitemap no longer contains /integrations/eliza (removed; 301 to /integrations/elizaos).
  • Founder Person schema cleaned — removed fabricated-looking givenName/familyName/alumniOf/dead LinkedIn URL.

[0.4.0] — 2026-06-28

  • Initial PyPI release.
  • OFAC sanctions screening (sanctions_check), transaction risk scoring (risk_score), Know-Your-Agent verification (kya_verify), dispute management (dispute_open).
  • MCP server, HTTP API, and CLI interfaces.
  • Free tier: 5 checks/day, no API key.
Assets 2
Loading

v0.4.0 — Compliance & verification toolkit for AI agents

v0.4.0 — Compliance & verification toolkit for AI agents

Choose a tag to compare

View all tags
@kindrat86 kindrat86 released this 28 Jun 19:08
v0.4.0
754f6b5

What's in this release

Compliance layer (real OFAC data):

  • sanctions_check — screen wallets/names/countries against 782 crypto addresses + 19,086 names + 16 embargoed jurisdictions
  • risk_score — pre-payment fraud scoring (allow/review/decline)
  • kya_verify — Know-Your-Agent trust scoring
  • dispute_open — transaction dispute registry

Verification toolkit:

  • Email: disposable verification inboxes (Mail.tm)
  • SMS: rentable phone numbers (mock/5sim/twilio)

Three surfaces: MCP server (10 tools), HTTP API (with hosted mode), CLI.

Data sources (all free, no API key):

  • vile/ofac-sdn-list GitHub releases (crypto addresses, daily refresh)
  • US Treasury OFAC sdn.csv (names + countries)

Hosted API mode: API-key auth, rate limiting, free-tier cap, audit log — all env-driven, off by default.

pip install agentmail[mcp]
agentmail sanctions --wallet 0x098B716B8Aaf21512996dC57EB0615e2383E2f96
# → flagged: OFAC_SDN wallet_exact

MIT licensed. Python 3.11+.

Loading