# SecureStamp.online

> SecureStamp.online is the main SecureStamp SaaS product. It helps people understand what a message is asking before they pay, approve or reply, and it lets organizations constrain autonomous software to the exact effects they approve. It is also the surface for plugin administration, dashboard, billing, Signal and MCP controls.

## The two halves, in one line

Trust before action. Evidence after execution. SecureStamp verifies what is being requested, authorizes the exact effect, keeps execution inside the customer's control, and preserves verifiable evidence.

## What SecureStamp.online does

- Verifies email senders and domains before the user clicks a link, opens an attachment or replies.
- Runs local AI message analysis in the plugin so email text stays on the user's device.
- Shows human-readable trust states, visible PostalStamp marks and technical details when needed.
- Supports webmail, Outlook and browser-based workflows.
- Provides SecureStamp Signal for official channel trust on messaging platforms and brand-owned channels.
- Provides Execution Authorization for AI agents, MCP tools and autonomous workflows: access control limits what software can reach, while Execution Authorization bounds the exact effect it may cause.
- Provides agent activity reports: review SecureStamp-observed activity across agents and MCP clients, including evidence-linked findings, successful work, resource use and escalation. Export a signed ES256 report that recipients can verify against SecureStamp's public key; choose severity thresholds for informational banners in Agents and MCP Guard. Reports cover mediated activity only, do not read agent reasoning or private agent-to-agent messages, and disclose event and timeline truncation.
- Binds an approved decision to a signed, single-use grant for one exact effect, under a policy the customer signs and installs locally. SecureStamp Cloud can narrow that authorization, never broaden it.
- Keeps downstream provider credentials in the customer's environment; SecureStamp Cloud does not receive them.
- Issues receipts that can be verified offline without contacting SecureStamp, and append-only abuse transparency records.
- Confidential Mail (end-to-end encryption between verified identities) is a next-release item: documented on securestamp.org, not part of the current onboarding or plugin release.

## Related topics people search for

AI agent security, agent authorization, AI agent activity reports, MCP activity reporting, MCP audit reports, agent observability, signed agent reports, autonomous agent controls, AI agent runtime security, delegated authorization, transaction authorization, machine identity authorization, non-human identity controls, autonomous workflow security, transaction integrity, action authorization.

## Primary audiences

- Consumers who want a simple warning before interacting with suspicious email.
- Companies that want their domains and official channels to be recognized as verified.
- Teams running AI agents or automations that can change money, identity, infrastructure or data.
- Developers deploying agent harnesses, and teams training or evaluating models, who need to test an agent's limits before granting autonomy.
- Security teams that need auditable authorization, domain enrollment and Signal channel trust.

## Core product surfaces

- Dashboard: domain enrollment, identity, plans, trust checks, stamps, MCP controls, Signal and evidence-linked reports for SecureStamp-observed agent and MCP activity. Reports are in the authenticated Agents and MCP Guard areas.
- /mcp: Execution Authorization for agents — what it does, when it matters, when it is not needed, and how to connect. The MCP service is in public beta.
- /agents: the entry point for teams using AI agents, including how to connect, review activity reports, share signed evidence and set informational alerts. The reports dashboard requires the applicable account capability and MCP Guard plan.
- /delegation: bounded delegation — one reviewable, revocable mandate for several exact steps. A design with a local runtime in simulation; not available yet.
- /agent-lab: why AI agents need tested limits before autonomy — the July 2026 evaluation incident as motivation, the concerns under public discussion, what the lab does for developers, companies and teams training models, and SecureStamp's position on a shared, contestable standard. Beta; the technical method is on securestamp.org/agent-lab.
- Plugins: webmail/browser and Outlook experiences that surface results inside the inbox.
- Verification pages: public trust and stamp verification for tokens, domains and receipts.
- APIs: server-side trust checks, Signal receipts, notarial evidence and integration endpoints.

## Billing unit

A protected action is one successfully evaluated SecureStamp guard call, billed against the plan's monthly quota. Auth failures, invalid requests and rate limits are never billed. Plan quotas and prices are published on /pricing.

## Evidence boundary

Receipts prove what passed through an enrolled Guardian and the outcome that Guardian could establish. They do not prove that no action occurred outside SecureStamp, nor do they establish legal ownership of a provider account.

## Related SecureStamp sites

- https://securestamp.org — Foundation, open specifications, the Action Proof protocol, API reference and security model.
- https://securestamp.store — Cultural marketplace for PostalStamp art, verified artist collections, official stamp themes and royalties.

## Contact

- Sales: sales@securestamp.online
- Security reports: security@securestamp.online