{"version":"2.1.251","anchor":"eval-authoring-guidance-now-requires-mocking-mcp-servers","canonical_anchor":"eval-authoring-guidance-now-requires-mocking-mcp-servers","heading":"Eval-authoring guidance now requires mocking MCP servers","tier":"notice","area":"Skills","scope":null,"heads_up":null,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.251\/e\/eval-authoring-guidance-now-requires-mocking-mcp-servers","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.251","markdown":"### Eval-authoring guidance now requires mocking MCP servers\n\nThe eval-writing skill now requires stubbing MCP calls so evals never hit real services.\n\n**What**\n\nThe built-in skill for writing evals gained a required step telling authors to stub every MCP tool call an eval can make, so evals never reach real MCP services.\n\n**Details**\n\n- each tool gets a mock file under `mocks\/<server>\/<tool>.md`\n- flows that make several calls can use a single `_server.md` agent mock with an `abort_when:` list\n\n**Evidence**\n\n`Step 3a \u2014 Mock the MCP servers`\n\n- Area: Skills\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5"}