Understand
List accessible projects, inspect project details and review recent build results.
AI agent access
Let your AI client work with P2HS
The P2HS MCP connector gives an authorized agent a structured way to understand projects, estimate builds, start work and follow progress—while Core keeps workspace access, billing and permissions authoritative.
Endpoint: https://mcp.p2hs.com/mcp
MCP / authorized session
→ list_projects
→ estimate_build
→ start_build
→ get_operation
The client can act on project context without receiving private Core credentials or bypassing P2HS authorization.
What it unlocks
A useful agent surface for the build lifecycle
Use an MCP-compatible client as a planning and delivery partner, with permissions that stay explicit and revocable.
Plan
Estimate the credits a build may use before any execution begins.
Build
Submit an authorized build with prompts and optional reference attachments.
Track
Poll durable operations and keep the client aligned with the current status.
Designed for control
The agent gets a grant. P2HS keeps the policy.
OAuth with PKCE, scoped workspace service accounts, selected-project access, expiry and revocation keep the connection bounded. Mutations use request keys so retries can remain safe and idempotent.
Project-scoped by default
Connections can start with selected-project access and expand only when the workspace owner allows it.
Billing-aware
Build estimates, credit packages and purchase flows are subject to Core authorization and workspace controls.
No private credentials in the client
The gateway talks to the dashboard; private Core hosts and shared administrator credentials stay server-side.
Start with the connection guide
See the endpoint, discovery documents, scopes and the first safe request flow.