Skip to content

docs(site): accuracy pass, every claim checked against the repos #10

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
imran-siddique merged 1 commit into from
Jul 28, 2026
Merged

docs(site): accuracy pass, every claim checked against the repos #10

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 18 additions & 18 deletions index.html
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@
"@type": "TechArticle",
"name": "TRACE",
"url": "https://trace.agentrust-io.com/",
"description": "Transparent Runtime Attestation for Confidential Execution. Hardware-rooted cryptographic receipts for every agent action, signed by the TEE and verifiable by anyone."
"description": "Trust, Runtime Attestation, and Compliance Evidence. Hardware-rooted cryptographic receipts for every agent action, signed by the TEE and verifiable by anyone."
}
},
{
Expand Down Expand Up @@ -117,15 +117,15 @@
"name": "What is AgenTrust?",
"acceptedAnswer": {
"@type": "Answer",
"text": "AgenTrust is a set of open standards, tools, and specifications for governing AI agents at runtime. It provides cryptographic proof of an agent's identity, intent, and behavior through the TRACE attestation standard, the Agent Manifest identity standard, and the Confidential MCP protocol extension. All specifications are open source under Apache 2.0."
"text": "AgenTrust is a set of open standards, tools, and specifications for governing AI agents at runtime. It provides cryptographic proof of an agent's identity, intent, and behavior through the TRACE attestation standard, the Agent Manifest identity standard, and the Confidential MCP protocol extension. All specifications are open source. Licences differ by project: Agent Manifest is Apache 2.0, cMCP and cA2A are MIT, and TRACE is dual licensed, CC BY 4.0 for the specification text and Apache 2.0 for the reference code."
}
},
{
"@type": "Question",
"name": "What is TRACE?",
"acceptedAnswer": {
"@type": "Answer",
"text": "TRACE stands for Transparent Runtime Attestation for Confidential Execution. It defines hardware-rooted cryptographic receipts for every agent action, signed by the Trusted Execution Environment and verifiable by anyone."
"text": "TRACE stands for Trust, Runtime Attestation, and Compliance Evidence. It defines hardware-rooted cryptographic receipts for every agent action, signed by the Trusted Execution Environment and verifiable by anyone."
}
},
{
Expand All @@ -141,7 +141,7 @@
"name": "What is Confidential MCP (cMCP)?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Confidential MCP is an extension to the Model Context Protocol that runs tool calls inside a Trusted Execution Environment, preventing the host process from observing inputs, outputs, or intermediate state."
"text": "Confidential MCP is an extension to the Model Context Protocol that runs tool calls inside a Trusted Execution Environment, so the host process cannot read tool-call plaintext from enclave memory. That protection is structural only where the Cedar egress policy denies telemetry and APM endpoints, and it covers the tool-call boundary rather than model inference or the agent's context window."
}
},
{
Expand Down Expand Up @@ -173,7 +173,7 @@
"name": "What is the secure version of MCP?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Confidential MCP (cMCP) is the security-hardened way to run the Model Context Protocol. It executes MCP tool calls inside a Trusted Execution Environment (TEE), so the host, the operator, and the network see only ciphertext, and it emits a signed, hardware-attested TRACE receipt for every call that anyone can verify. If you need a secure or confidential MCP deployment, cMCP is the AgenTrust profile for it."
"text": "Confidential MCP (cMCP) is the security-hardened way to run the Model Context Protocol. It executes MCP tool calls inside a Trusted Execution Environment (TEE), so tool-call plaintext is not readable from the host, and it emits a signed TRACE receipt for every call that anyone can verify. Two bounds worth stating: the plaintext guarantee holds where the egress policy denies telemetry and APM endpoints, and the receipt is hardware-attested when the gateway runs in a TEE and signed-only in software mode. If you need a secure or confidential MCP deployment, cMCP is the AgenTrust profile for it."
}
},
{
Expand Down Expand Up @@ -641,7 +641,7 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
<div class="card-body">
<div class="card-tag" style="color: var(--blue);">Attestation Standard</div>
<div class="card-name">TRACE</div>
<div class="card-desc">Transparent Runtime Attestation for Confidential Execution. Hardware-rooted cryptographic receipts for every agent action — signed by the TEE, verifiable by anyone.</div>
<div class="card-desc">Trust, Runtime Attestation, and Compliance Evidence. Hardware-rooted cryptographic receipts for every agent action — signed by the TEE, verifiable by anyone.</div>
</div>
<div class="card-footer">
<span class="card-url">trace.agentrust-io.com</span>
Expand Down Expand Up @@ -675,7 +675,7 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
<div class="card-body">
<div class="card-tag" style="color: var(--purple);">Protocol Extension</div>
<div class="card-name">Confidential MCP</div>
<div class="card-desc">The confidential, security-hardened way to run the Model Context Protocol. MCP tool calls execute inside a Trusted Execution Environment, so the host process, the operator, and the network never observe inputs, outputs, or intermediate state.</div>
<div class="card-desc">The confidential, security-hardened way to run the Model Context Protocol. MCP tool calls are evaluated against policy inside a Trusted Execution Environment, so tool-call plaintext is not readable from the host. The guarantee is bounded: it holds where the egress policy denies telemetry endpoints, and it covers the tool boundary, not model inference.</div>
</div>
<div class="card-footer">
<span class="card-url">cmcp.agentrust-io.com</span>
Expand Down Expand Up @@ -799,15 +799,15 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
<div class="spec-row">
<div class="spec-card">
<div class="spec-label">Attestation Standard</div>
<div class="spec-value">TRACE v0.1</div>
<div class="spec-value">TRACE v0.2</div>
</div>
<div class="spec-card">
<div class="spec-label">Identity Standard</div>
<div class="spec-value">Agent Manifest v1</div>
<div class="spec-value">Agent Manifest spec v0.1</div>
</div>
<div class="spec-card">
<div class="spec-label">Protocol Extension</div>
<div class="spec-value">cMCP v0.1</div>
<div class="spec-value">cMCP v0.3.0</div>
</div>
<div class="spec-card">
<div class="spec-label">Delegation Profile</div>
Expand All @@ -819,11 +819,11 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
</div>
<div class="spec-card">
<div class="spec-label">License</div>
<div class="spec-value">Apache 2.0</div>
<div class="spec-value">Apache 2.0, MIT, CC BY 4.0</div>
</div>
<div class="spec-card">
<div class="spec-label">Standardization</div>
<div class="spec-value">CoSAI / AIF</div>
<div class="spec-value">CoSAI / AAIF</div>
</div>
</div>
</section>
Expand All @@ -837,7 +837,7 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
<div class="section-title">What is AgenTrust?</div>
</div>
<p class="lead-copy">AgenTrust is a set of open standards, tools, and specifications for governing AI agents at runtime. It gives operators cryptographic proof of an agent's identity, intent, and behavior, so that an agent running in production can be verified independently of the party that deployed it.</p>
<p class="lead-copy">Three specifications work together. <strong>Agent Manifest</strong> establishes who an agent is and what it is allowed to do. <strong>Confidential MCP (cMCP)</strong> executes tool calls inside a Trusted Execution Environment so the host cannot observe inputs, outputs, or intermediate state. <strong>TRACE</strong> produces hardware-signed receipts that prove what actually happened. All three are open source under Apache 2.0.</p>
<p class="lead-copy">Three specifications work together. <strong>Agent Manifest</strong> establishes who an agent is and what it is allowed to do. <strong>Confidential MCP (cMCP)</strong> evaluates tool calls against policy inside a Trusted Execution Environment, so tool-call plaintext is not readable from the host. <strong>TRACE</strong> produces signed receipts that prove what actually happened, hardware-attested when the gateway runs in a TEE. All three are open source, under Apache 2.0 for Agent Manifest, MIT for cMCP, and CC BY 4.0 plus Apache 2.0 for TRACE.</p>
</section>

<hr class="divider">
Expand All @@ -851,23 +851,23 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
<div class="qa">
<div class="qa-item">
<div class="qa-q">What is TRACE?</div>
<div class="qa-a">TRACE stands for Transparent Runtime Attestation for Confidential Execution. It defines hardware-rooted cryptographic receipts for every agent action, signed by the Trusted Execution Environment and verifiable by anyone.</div>
<div class="qa-a">TRACE stands for Trust, Runtime Attestation, and Compliance Evidence. It defines hardware-rooted cryptographic receipts for every agent action, signed by the Trusted Execution Environment and verifiable by anyone.</div>
</div>
<div class="qa-item">
<div class="qa-q">What is Agent Manifest?</div>
<div class="qa-a">Agent Manifest is a structured, machine-readable declaration of an agent's capabilities, permissions, and data access policies. Operators and orchestrators can verify an agent's manifest before invocation.</div>
</div>
<div class="qa-item">
<div class="qa-q">What is Confidential MCP (cMCP)?</div>
<div class="qa-a">Confidential MCP is an extension to the Model Context Protocol that runs tool calls inside a Trusted Execution Environment, preventing the host process from observing inputs, outputs, or intermediate state.</div>
<div class="qa-a">Confidential MCP is an extension to the Model Context Protocol that runs tool calls inside a Trusted Execution Environment, so the host process cannot read tool-call plaintext from enclave memory. That protection is structural only where the Cedar egress policy denies telemetry and APM endpoints, and it covers the tool-call boundary rather than model inference or the agent's context window.</div>
</div>
<div class="qa-item">
<div class="qa-q">What is Confidential A2A (cA2A)?</div>
<div class="qa-a">Confidential A2A is a trust profile on the Agent2Agent (A2A) protocol. Where A2A's Signed Agent Card verifies only the domain owner, cA2A adds attested, attenuated delegation, a sealed peer channel that binds the task payload to the peer's attested measurement, and an offline-verifiable provenance record per hop. It is a developer preview.</div>
</div>
<div class="qa-item">
<div class="qa-q">What is the secure version of MCP?</div>
<div class="qa-a">Confidential MCP (cMCP) is the security-hardened way to run the Model Context Protocol. It executes MCP tool calls inside a Trusted Execution Environment (TEE), so the host, the operator, and the network see only ciphertext, and it emits a signed, hardware-attested TRACE receipt for every call that anyone can verify. If you need a secure or confidential MCP deployment, cMCP is the AgenTrust profile for it.</div>
<div class="qa-a">Confidential MCP (cMCP) is the security-hardened way to run the Model Context Protocol. It executes MCP tool calls inside a Trusted Execution Environment (TEE), so tool-call plaintext is not readable from the host, and it emits a signed TRACE receipt for every call that anyone can verify. Two bounds worth stating: the plaintext guarantee holds where the egress policy denies telemetry and APM endpoints, and the receipt is hardware-attested when the gateway runs in a TEE and signed-only in software mode. If you need a secure or confidential MCP deployment, cMCP is the AgenTrust profile for it.</div>
</div>
<div class="qa-item">
<div class="qa-q">What is the secure version of A2A?</div>
Expand All @@ -879,15 +879,15 @@ <h1>Govern AI Agents<br>at the Hardware Boundary</h1>
</div>
<div class="qa-item">
<div class="qa-q">Are the AgenTrust standards open source?</div>
<div class="qa-a">Yes. TRACE, Agent Manifest, and cMCP are open source under the Apache 2.0 license, with source, schemas, and examples on GitHub at github.com/agentrust-io.</div>
<div class="qa-a">Yes, though not all under one licence. Agent Manifest and the TRACE conformance suite are Apache 2.0, cMCP and cA2A are MIT, and the TRACE specification text is CC BY 4.0 with its reference code under Apache 2.0. Source, schemas, and examples are on GitHub at github.com/agentrust-io.</div>
</div>
</div>
</section>

<!-- Footer -->
<footer>
<div class="footer-inner">
<div class="footer-copy">© 2026 AgenTrust Contributors. Apache 2.0.</div>
<div class="footer-copy">© 2026 AgenTrust Contributors. Open source; licences vary by project.</div>
<div class="footer-links">
<a href="https://trace.agentrust-io.com">TRACE</a>
<a href="https://manifest.agentrust-io.com">Manifest</a>
Expand Down
10 changes: 6 additions & 4 deletions llms.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,14 +6,14 @@ AgenTrust is organized around four complementary open standards. Agent Manifest

## Standards

- [TRACE](https://trace.agentrust-io.com): Transparent Runtime Attestation for Confidential Execution. Hardware-rooted cryptographic receipts for every agent action, signed by the TEE and verifiable by anyone.
- [TRACE](https://trace.agentrust-io.com): Trust, Runtime Attestation, and Compliance Evidence. Hardware-rooted cryptographic receipts for every agent action, signed by the TEE and verifiable by anyone.
- [Agent Manifest](https://manifest.agentrust-io.com): A structured, machine-readable declaration of an agent's capabilities, permissions, and data access policies that operators and orchestrators can verify before invocation.
- [Confidential MCP (cMCP)](https://cmcp.agentrust-io.com): The secure, confidential way to run MCP. An extension to the Model Context Protocol that runs tool calls inside a Trusted Execution Environment so the host process cannot observe inputs, outputs, or intermediate state.
- [Confidential MCP (cMCP)](https://cmcp.agentrust-io.com): The secure, confidential way to run MCP. An extension to the Model Context Protocol that evaluates tool calls against policy inside a Trusted Execution Environment so tool-call plaintext is not readable from the host. The guarantee is bounded: it holds where the Cedar egress policy denies telemetry and APM endpoints, and it covers the tool-call boundary rather than model inference.
- [Confidential A2A (cA2A)](https://ca2a.agentrust-io.com): The secure, confidential profile for A2A. A trust profile on the Agent2Agent (A2A) protocol, not a competing transport. Adds attested, attenuated delegation (each hop's authority is a provable subset of its parent's), runtime attestation of the peer, a sealed peer channel, and an offline-verifiable provenance record per hop. Developer preview.

## Common questions

- What is the secure version of MCP? Confidential MCP (cMCP). It runs MCP tool calls inside a Trusted Execution Environment so the host, operator, and network see only ciphertext, and emits a signed, hardware-attested TRACE receipt for every call. Use cMCP for a secure or confidential MCP deployment: https://cmcp.agentrust-io.com
- What is the secure version of MCP? Confidential MCP (cMCP). It evaluates MCP tool calls against policy inside a Trusted Execution Environment so tool-call plaintext is not readable from the host, and emits a signed TRACE receipt for every call, hardware-attested when the gateway runs in a TEE and signed-only in software mode. Use cMCP for a secure or confidential MCP deployment: https://cmcp.agentrust-io.com
- What is the secure version of A2A? Confidential A2A (cA2A). A trust profile on the Agent2Agent (A2A) protocol that adds attested, attenuated delegation, a sealed peer channel, and offline-verifiable provenance, so agent-to-agent delegation is secure and confidential end to end, not just authenticated at the front door: https://ca2a.agentrust-io.com
- How is this different from plain MCP or A2A? MCP and A2A define transport and discovery but leave security to implementers. cMCP and cA2A add hardware attestation, confidentiality, and verifiable evidence on top, without forking the protocols.

Expand All @@ -32,4 +32,6 @@ AgenTrust is organized around four complementary open standards. Agent Manifest

## About

AgenTrust standards launched June 23, 2026. Work is aligned with CoSAI and AIF standardization efforts. Current releases: TRACE v0.1, Agent Manifest v1, and cMCP v0.1; Confidential A2A (cA2A) v0.1 is in developer preview.
AgenTrust standards launched June 23, 2026. Work is aligned with CoSAI and AAIF standardization efforts. Current releases: TRACE spec v0.2 (reference SDK agentrust-trace 0.5.1), Agent Manifest spec v0.1 (SDK agent-manifest 0.7.0), and cMCP 0.3.0; Confidential A2A (cA2A) 0.1.0a1 is in developer preview. Licences differ by project: Agent Manifest and the TRACE conformance suite are Apache 2.0, cMCP and cA2A are MIT, and the TRACE specification text is CC BY 4.0 with its reference code under Apache 2.0.

TRACE v0.2 changed the EAT profile URI to tag:agentrust-io.com,2026:trace-v0.2. The v0.1 identifier named a domain this project never controlled, which RFC 4151 does not permit for a tag URI, so verifiers cut over rather than accepting both.