DUBLIN, California — September 18, 2026 — Safeguard's connector is now available in ChatGPT, alongside its existing listing in the Claude directory. Developers and security teams can now ask about their own software supply chain from inside the assistant they already have open, and get an answer drawn from their live Safeguard account rather than from the model's memory.
The connector is discoverable from within the assistant itself. Users can search for Safeguard, connect once, and from then on ask questions in the same conversation where they are already reading code, drafting a fix, or preparing a release.
What it answers
The connector exposes read-only queries over the estate the signed-in user already has access to:
- Products and projects — what is being scanned, its scan status, and the vulnerability counts that roll up to it.
- Vulnerabilities — filtered by severity, ecosystem, CVSS or exploit status, and the full record for a given CVE including its references.
- Findings — the individual occurrences behind those counts, for a specific project or SBOM.
Answers name the CVE, the affected package and version, and the fix version where one is known. Questions like "which of my projects are affected by this CVE" and "what is the fix version for this package" stop being a context switch into a dashboard.
Scope and access
Three properties are worth stating plainly, because this is a security product and the way an assistant reaches it matters as much as what it returns.
Sign-in is required. The connector authenticates over OAuth against the user's own Safeguard account. It holds no shared credential and grants no anonymous access; an unauthenticated request to the MCP endpoint is refused before any tool is listed.
Scope comes from the session, not the prompt. Which tenant, organization and product a query covers is decided by the authenticated session, and applied automatically. It is not something the conversation can ask for and not something a prompt can widen.
The tools are read-only. The connector queries; it does not modify configuration, start scans, or connect repositories.
Availability
The connector is available now in ChatGPT and in the Claude directory. There is no separate purchase: it reads the account a user already has, including the free tier.
- ChatGPT: https://chatgpt.com/plugins/plugin_asdk_app_6a26c2110a40819196c8c8fcc0d29ae9?open_in_app
- Claude directory: https://claude.ai/directory/safeguard
- MCP endpoint, for clients that connect directly:
https://mcp.safeguard.sh/mcp
Safeguard's MCP server also backs its integrations with coding agents including Claude Code and Cursor, documented at https://safeguard.sh/products/mcp-server.
About Safeguard
Safeguard is an AI-native software supply chain security platform. It maintains a registry of curated zero-CVE components across ten ecosystems, discovers Zero Days through reachability and taint analysis deep in the dependency tree, and authors remediation directly against the affected code. Safeguard is headquartered in Dublin, California. Pricing is public and self-serve from $0 at https://safeguard.sh/pricing.
Safeguard is not affiliated with, endorsed by, or a partner of OpenAI or Anthropic. ChatGPT is a trademark of OpenAI. Claude is a trademark of Anthropic.
