Skip to main content
Every data tool takes an optional project (a project UUID) and an optional env, plus service / key where applicable. When project is omitted it is resolved from .nexus-mcp.json (or the home registry); when env is omitted the backend uses your personal dev environment. Read tools are annotated readOnlyHint; reveal and write tools are annotated destructiveHint, so the host client asks you to confirm before running them.

Authentication tools

Discovery

Read tools

Reveal and write tools

Error messages

Backend errors are translated to clear, actionable messages, for example:
  • mcp_project_required -> “No project specified. Pass project:<uuid>, add a .nexus-mcp.json, or run list_projects.”
  • mcp_not_member -> “You are not a member of that project.”
  • mcp_env_not_allowed -> “That environment is not in your MCP scope (baseline personal env, or your grant).”
  • mcp_reveal_prod_forbidden -> “Revealing secrets in production is never permitted.”
  • mcp_write_not_allowed -> “Writing is not permitted by your MCP grant (this grant is read-only).”
  • mcp_create_not_allowed -> “Creating new items is not permitted by your MCP grant.”
  • mcp_delete_not_allowed -> “Delete is only allowed in your personal dev env, not on shared environments.”
All enforcement happens on the backend; the server only forwards your call and translates the result.