Agentdataboundary Mcp
Emit permission boundary receipts so ChatGPT-class agents prove they stayed inside approved data access limits.
Overview
AgentData Boundary MCP is a MCP server for the Ship phase that issues permission boundary receipts for ChatGPT agents and data-governance audits.
What is this MCP server?
- Permission boundary receipts tailored for ChatGPT-style agents
- Remote MCP with streamable-http and Bearer authorization
- Data-governance focus with publisher tags: data-governance, permissions, remote-mcp
- Server card and GitHub source (clauxel/agent-data-boundary-mcp), version 1.0.0
- Evidence-oriented design for audits—not just static policy docs in a repo
- Version 1.0.0 in the MCP registry
- One streamable-http remote at agentdataboundary.clauxel.com/mcp
- Three publisher tags: data-governance, permissions, remote-mcp
What problem does it solve?
You cannot easily show which data an agent was allowed to access versus what it actually attempted during a run.
Who is it for?
Builders shipping agents that touch PII or multi-tenant data who need lightweight permission evidence in MCP workflows.
Skip if: Static sites with no agent data access, or enterprises that require a full GRC platform instead of receipt-oriented MCP tooling.
What do I get? / Deliverables
After install you get machine-readable boundary receipts your agent can reference during reviews and policy checks.
- Permission boundary receipts per agent run or check
- Reviewable evidence aligned to data-governance reviews
- MCP-callable boundary context for security-minded prompts
Recommended MCP Servers
Journey fit
Data and permission boundaries are enforced before and after launch; the canonical shelf is ship/security because that is when you gate risky agent access patterns. Security subphase covers governance receipts that document what an agent was allowed to touch versus what it actually requested.
How it compares
Governance receipt MCP for agent permissions, not a secrets vault or WAF.
Common Questions / FAQ
Who is AgentData Boundary MCP for?
It is for indie developers and small teams running data-connected agents who need documented permission boundaries.
When should I use AgentData Boundary MCP?
Use it while hardening agent integrations before launch, during security review, or when updating data policies in production.
How do I add AgentData Boundary MCP to my agent?
Add the remote URL https://agentdataboundary.clauxel.com/mcp to your MCP client, configure the Authorization header with a Bearer token from Clauxel, and enable streamable-http per the registry entry.