34578d39c6
Move src/ → apps/server/ and packages/web/ → apps/web/ to adopt standard monorepo conventions (apps/ for runnable apps, packages/ for reusable libraries). Update all config files, shared package imports, test fixtures, and documentation to reflect new paths. Key fixes: - Update workspace config to ["apps/*", "packages/*"] - Update tsconfig.json rootDir/include for apps/server/ - Add apps/web/** to vitest exclude list - Update drizzle.config.ts schema path - Fix ensure-schema.ts migration path detection (3 levels up in dev, 2 levels up in dist) - Fix tests/integration/cli-server.test.ts import paths - Update packages/shared imports to apps/server/ paths - Update all docs/ files with new paths
114 lines
3.4 KiB
TypeScript
114 lines
3.4 KiB
TypeScript
/**
|
|
* Minimal Cost Test Prompts
|
|
*
|
|
* Carefully crafted prompts designed to minimize token usage while
|
|
* testing specific CLI behaviors. Each prompt aims for the smallest
|
|
* possible API cost while still exercising the target functionality.
|
|
*
|
|
* Cost estimates assume Claude Sonnet pricing (~$3/M input, $15/M output).
|
|
*/
|
|
|
|
export const MINIMAL_PROMPTS = {
|
|
/**
|
|
* ~$0.01 - Cheapest done response
|
|
* Tests: basic spawn → completion flow, status parsing
|
|
*/
|
|
done: `Output exactly this JSON with no other text:
|
|
{"status":"done","result":"ok"}`,
|
|
|
|
/**
|
|
* ~$0.01 - Cheapest questions response
|
|
* Tests: waiting_for_input status, questions array parsing
|
|
*/
|
|
questions: `Output exactly this JSON with no other text:
|
|
{"status":"questions","questions":[{"id":"q1","question":"What is your name?"}]}`,
|
|
|
|
/**
|
|
* ~$0.03 - Slow task for timing tests
|
|
* Tests: streaming during long-running task, crash recovery
|
|
* Note: Agent may not actually wait 30 seconds, but will produce delayed output
|
|
*/
|
|
slow: `Think through a simple problem step by step, counting from 1 to 10 slowly, then output:
|
|
{"status":"done","result":"counted to 10"}`,
|
|
|
|
/**
|
|
* ~$0.02 - Produces text deltas for streaming tests
|
|
* Tests: text_delta event parsing, output buffering
|
|
*/
|
|
streaming: `Count from 1 to 5, outputting each number, then output:
|
|
{"status":"done","result":"counted"}`,
|
|
|
|
/**
|
|
* ~$0.03 - Deliberately produces non-JSON first
|
|
* Tests: schema validation failure, retry logic
|
|
*/
|
|
badThenGood: `First say "thinking..." on its own line, then output:
|
|
{"status":"done","result":"fixed"}`,
|
|
|
|
/**
|
|
* ~$0.02 - Multiple questions
|
|
* Tests: questions array with multiple items
|
|
*/
|
|
multipleQuestions: `Output exactly this JSON with no other text:
|
|
{"status":"questions","questions":[{"id":"q1","question":"First question?"},{"id":"q2","question":"Second question?"}]}`,
|
|
|
|
/**
|
|
* ~$0.01 - Error signal
|
|
* Tests: error status handling
|
|
*/
|
|
error: `Output exactly this JSON with no other text:
|
|
{"status":"error","error":"Test error message"}`,
|
|
|
|
/**
|
|
* ~$0.02 - Answer continuation
|
|
* Tests: session resume with answers
|
|
*/
|
|
answerContinuation: (answers: Record<string, string>): string => {
|
|
const answerLines = Object.entries(answers)
|
|
.map(([id, answer]) => `${id}: ${answer}`)
|
|
.join('\n');
|
|
return `I received your answers:
|
|
${answerLines}
|
|
|
|
Now complete the task by outputting:
|
|
{"status":"done","result":"completed with answers"}`;
|
|
},
|
|
|
|
/**
|
|
* ~$0.02 - Context complete for discuss mode
|
|
* Tests: discuss mode output handling (now uses universal done signal)
|
|
*/
|
|
discussComplete: `Output exactly this JSON with no other text:
|
|
{"status":"done"}`,
|
|
|
|
/**
|
|
* ~$0.02 - Plan complete
|
|
* Tests: plan mode output handling (now uses universal done signal)
|
|
*/
|
|
planComplete: `Output exactly this JSON with no other text:
|
|
{"status":"done"}`,
|
|
|
|
/**
|
|
* ~$0.02 - Detail complete
|
|
* Tests: detail mode output handling (now uses universal done signal)
|
|
*/
|
|
detailComplete: `Output exactly this JSON with no other text:
|
|
{"status":"done"}`,
|
|
} as const;
|
|
|
|
/**
|
|
* Prompts specifically for Codex provider testing.
|
|
* Codex may have different output format requirements.
|
|
*/
|
|
export const CODEX_PROMPTS = {
|
|
/**
|
|
* Basic completion for Codex
|
|
*/
|
|
done: `Complete this simple task: output "done" and finish.`,
|
|
|
|
/**
|
|
* Produces streaming output
|
|
*/
|
|
streaming: `Count from 1 to 5, saying each number aloud, then say "finished".`,
|
|
} as const;
|