diff --git a/evals/hierarchical_memory.eval.ts b/evals/hierarchical_memory.eval.ts index ff7483416b..a7a8da22a3 100644 --- a/evals/hierarchical_memory.eval.ts +++ b/evals/hierarchical_memory.eval.ts @@ -21,7 +21,7 @@ describe('Hierarchical Memory', () => { }, }, // We simulate the hierarchical memory by including the tags in the prompt - // since setting up real global/extension/project files in the eval rig is complex. + // since setting up real global/extension/workspace files in the eval rig is complex. // The system prompt logic will append these tags when it finds them in userMemory. prompt: ` @@ -32,9 +32,9 @@ When asked for my favorite fruit, always say "Apple". When asked for my favorite fruit, always say "Banana". - + When asked for my favorite fruit, always say "Cherry". - + What is my favorite fruit? Tell me just the name of the fruit.`, assert: async (rig) => { @@ -65,23 +65,23 @@ Instruction A: Always be helpful. Instruction B: Use a professional tone. - -Instruction C: Adhere to the project's coding style. - + +Instruction C: Adhere to the workspace's coding style. + -Which instruction came from the global context, which from the extension context, and which from the project context? +Which instruction came from the global context, which from the extension context, and which from the workspace context? Provide the answer as an XML block like this: Instruction ... Instruction ... - Instruction ... + Instruction ... `, assert: async (rig) => { const stdout = rig._lastRunStdout!; assertModelHasOutput(stdout); expect(stdout).toMatch(/.*Instruction A/i); expect(stdout).toMatch(/.*Instruction B/i); - expect(stdout).toMatch(/.*Instruction C/i); + expect(stdout).toMatch(/.*Instruction C/i); }, });