DUBLIN, Calif. — July 8, 2026 — Safeguard today announced agent-native onboarding, payments, and procurement for its software supply chain security platform. Twelve new tools in the company's remote MCP (Model Context Protocol) server allow any connected AI assistant — including Claude, ChatGPT, Microsoft Copilot, and AI-native IDEs such as Cursor — to carry out the complete customer journey: create the Safeguard account and workspace itself through a dedicated pre-authentication endpoint, sign the session in without a browser, discover the pricing plans offered with correct regional pricing, compare tiers and the features each unlocks, initiate a Stripe Checkout session, verify after payment that the invoice settled and the purchased tier's features activated, and follow the workspace's notification inbox as results arrive.
The release extends Safeguard's MCP server — which already exposes more than 650 governed tools for vulnerability management, SBOM analysis, compliance, and AI-powered remediation — into the commercial layer of the platform. An agent that discovers a team's scans are paused on a usage limit can now, in the same conversation, identify the plan that lifts the limit, produce a secure payment link, confirm the payment landed, and resume the paused work.
Pricing discovery was designed for how agents hold information about their users. Rather than requiring a clean country code, the plan-listing tool accepts every location signal an agent knows — country, country name, locale, timezone, continent or country group, city or district, currency, or IP address — and resolves them to the correct pricing region entirely on the server, explaining in its response which signal decided the region. Plans carry both a tier (PRO, MAX, 5X, 20X, ENTERPRISE) and an audience category spanning individual, team, and enterprise buyers, and a companion entitlements tool reports the feature flags a subscription has enabled, so an assistant can answer "what does my plan include?" from live platform state.
Payment itself deliberately stays out of the model's hands. The purchase tool returns a Stripe-hosted checkout URL — the same hardened payment page used by Safeguard's web application — so card data never passes through the MCP server, the agent, or the model. Purchases are structurally scoped to the authenticated workspace: the tools accept no tenant parameter, making cross-tenant purchase or invoice access impossible by construction.
Consistent with the rest of the platform's SOC 2-aligned posture, every one of the new tools is governed by Safeguard's per-tool feature-flag system: the self-serve set is enabled by default so a newly created workspace is immediately operational, and administrators can disable any individual tool per tenant — including checkout initiation on its own. Every billing and account call, including read-only pricing and invoice queries, emits a structured audit record with tenant, user, action, and outcome, complementing the durable money-event audit trail written by the platform's payment activation flow.
"Every SaaS vendor is about to discover that their next buyer might be an agent acting for a team that's already decided," said Hritik Kumar Sharma, Founder and CEO of Safeguard. "We wanted the first agent-native purchase experience in security to be done right: real regional pricing instead of guesses, tiers an agent can reason about, payment kept on Stripe's hardened checkout rather than in a model's context window, and every step flagged, scoped, and audited. The same assistant that finds your risk can now procure the capacity to fix it — without ever holding a card number."
The billing and procurement tools are available today on Safeguard's MCP server at mcp.safeguard.sh. Documentation, including a dedicated guide for agents and LLMs using the platform, is available at https://docs.safeguard.sh.
About Safeguard
Safeguard is the software supply chain security platform that fuses multiple scanners, a security-only AI model lineup (Griffin · Eagle · Lion), and reachability-aware reasoning to find what pattern scanners miss — from CVEs to candidate zero-days — and to ship the fix with cited reasoning. Learn more at https://safeguard.sh.