Terms of use

Effective date: 2026-04-27. Minimum-viable summary — full terms live at the URL below and are the source of truth.

What AgentReserve is

AgentReserve scores the public security posture of MCP (Model Context Protocol) servers. Scoring is automated, fact- and heuristic-based, and intentionally passive — we never invoke a tool that an MCP server advertises. Scores are a triage signal, not a certification.

Acceptable use

• You may scan any MCP server you operate or are authorized to assess. You may also scan publicly-advertised servers for research or buyer-side diligence.
• You may not use the API or webhooks to denial-of-service a third party, evade rate limits, or republish bulk reports in a way that misrepresents the source.
• Public reports may be linked, embedded, and quoted with attribution to AgentReserve.

No warranty

Reports are provided “as is” without warranty of any kind. Scoring heuristics are derived from public protocol metadata and may produce false positives or false negatives. If you believe a report is wrong, file a dispute via the link on the report itself; we will mark the report as under review while we investigate.

API contract

The /api/v1/* surface is a stable public contract. We will not break a v1 response shape without shipping /api/v2/* in parallel and offering a deprecation window. See the API documentation for the canonical schemas.

Account termination

We may revoke API keys, disable webhooks, or close accounts that violate these terms or the acceptable-use policy. You may close your own account by emailing support@agentreserve.dev.

Contact

Questions: legal@agentreserve.dev. Full terms (placeholder URL until the legal team publishes the long form): agentreserve.dev/legal/terms.