菜单
Create job files in the Commons monorepo. Use when the user wants to create a new job, backfill script, cache job, export job, sync job, bulk operation, cron task, or any background task that runs via HTTP POST endpoint.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/jakewaldrip/.dotfiles --skill create-job命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
| name | create-job |
| description | Create job files in the Commons monorepo. Use when the user wants to create a new job, backfill script, cache job, export job, sync job, bulk operation, cron task, or any background task that runs via HTTP POST endpoint. |
Create background jobs that run as HTTP POST endpoints in the Commons worker.
commons-packages/backend/src/jobs/ are auto-discovered by filenameHTTP POST /jobs/<filename-without-extension>script-execute-handler.ts dynamically require()s the file and wraps it in BatchJobRun.trap() for lifecycle tracking/scripts/<slug> and /cron/<slug> also work but /jobs/ is preferredcommons-packages/backend/src/jobs/{kebab-case-name}.ts
Tests: commons-packages/backend/src/jobs/__tests__/{kebab-case-name}.spec.ts
Shared helpers: commons-packages/backend/src/jobs/shared/ or commons-packages/backend/src/jobs/helpers/
| Category | Prefix | Use Case |
|---|
| Backfill | backfill- | Populate or fix existing data |
| Cache/ETL | cache- | Sync external data (BigQuery, APIs) into local tables |
| Export | export- | Generate files (CSV, PDF) and upload to GCS |
| Bulk operation | bulk- | Batch create/delete/update operations |
| Sync | sync- | Two-way data synchronization |
| Create | create- | Provision new records |
| Delete | delete- | Remove or soft-delete records |
| Update | update- | Modify existing records |
| Start workflow | start- | Trigger workflow creation for eligible patients |
| Cleanup | cleanup- | Fix data inconsistencies, remove duplicates |
| Cancel | cancel- | Cancel workflows or scheduled items |
From commons-packages/backend/src/jobs/helpers.ts:
// The canonical handler type — all jobs should conform to this
type ScriptHandler = (req: express.Request) => Promise<JSONData | undefined | void>;
// Throw to return a specific HTTP status code (e.g., 400 for bad input)
class StatusError extends Error { status: number; data: JSONData; }
// Subclass of StatusError — job is marked successful even when thrown
class AcceptableError extends StatusError {}
import type { Request } from 'express';
import { z } from 'zod';
import logger from '@commons/backend/lib/logging.js';
import type { ScriptHandler } from '@commons/backend/jobs/helpers.js';
import { StatusError } from '@commons/backend/jobs/helpers.js';
const TAG = 'job-name';
const RequestBody = z.object({
exampleParam: z.string(),
optionalParam: z.number().optional().default(100),
});
// Named export for testability
export const jobName = async (body: unknown) => {
const bodyResult = RequestBody.safeParse(body);
if (!bodyResult.success) {
logger.error(bodyResult.error, { tag: TAG });
throw new StatusError(400, 'Invalid request body');
}
const { exampleParam, optionalParam } = bodyResult.data;
logger.log(`Starting job with param=${exampleParam}`, TAG);
// ... business logic ...
const result = { processed: 0, succeeded: 0, failed: 0 };
logger.log('Job complete', TAG, result);
return result;
};
// Default export — thin wrapper required by the job handler
const main: ScriptHandler = async (req) => jobName(req.body);
export default main;
import type { Request } from 'express';
import { z } from 'zod';
import logger from '@commons/backend/lib/logging.js';
import type { ScriptHandler } from '@commons/backend/jobs/helpers.js';
import { StatusError } from '@commons/backend/jobs/helpers.js';
import knex from '@commons/backend/lib/knex.js';
const TAG = 'backfill-entity-name';
const RequestBody = z.object({
batchSize: z.number().optional().default(1000),
limit: z.number().optional(),
isDryRun: z.boolean().optional().default(false),
});
export const backfillEntityName = async (body: unknown) => {
const bodyResult = RequestBody.safeParse(body);
if (!bodyResult.success) {
logger.error(bodyResult.error, { tag: TAG });
throw new StatusError(400, 'Invalid request body');
}
const { batchSize, limit, isDryRun } = bodyResult.data;
let totalProcessed = 0;
while (limit === undefined || totalProcessed < limit) {
const currentBatchSize = limit !== undefined
? Math.min(batchSize, limit - totalProcessed)
: batchSize;
const batch = await knex('table_name')
.select('*')
.whereNull('targetColumn')
.limit(currentBatchSize);
if (batch.length === 0) break;
if (!isDryRun) {
// ... perform writes ...
}
totalProcessed += batch.length;
logger.log(`Processed ${totalProcessed} records`, TAG, { isDryRun });
}
return { isDryRun, batchSize, limit, totalProcessed };
};
const main: ScriptHandler = async (req) => backfillEntityName(req.body);
export default main;
async (req: Request) => Promise<JSONData | undefined | void>ScriptHandler type from @commons/backend/jobs/helpers.js for type annotationsafeParse; throw new StatusError(400, ...) on failurelogger calls: const TAG = 'job-file-name';@commons/backend/lib/logging.js{ processed, succeeded, failed } or similarisDryRun param and skip writes when truebatchSize + limit params with a while loop or chunked iteration.js import extensions — use .js extensions in import paths (TypeScript with ESM resolution)jsonSchema — use Zod for request body validation, not JSON Schemajest.mock(), call the named export directly@commons/backend/__tests__/helpers/factories/, call the named exportconst buildRequest = (body: any) => ({ body }) as any; for calling the default exportimport { jobName } from '@commons/backend/jobs/job-name.js';
describe('jobName', () => {
it('processes records correctly', async () => {
// Setup with factories or mocks
const result = await jobName({ exampleParam: 'value' });
expect(result).toEqual({ processed: 1, succeeded: 1, failed: 0 });
});
it('rejects invalid input', async () => {
await expect(jobName({})).rejects.toThrow();
});
});
# Local
curl -X POST "http://localhost:8080/jobs/your-job-name" \
-H "Content-Type: application/json" \
-d '{"exampleParam": "value"}'
# Staging/Production (requires API Caller permission)
curl -X POST "https://commons-staging.cityblock.com/jobs/your-job-name" \
-H "Authorization: Bearer $(gcloud auth print-identity-token)" \
-H "Content-Type: application/json" \
-d '{"exampleParam": "value"}'
commons-packages/backend/src/handlers/script-execute-handler.tscommons-packages/backend/src/jobs/helpers.tscommons-packages/backend/src/jobs/shared/commons-packages/backend/src/jobs/helpers/commons-packages/backend/src/jobs/README.md基于 SOC 职业分类
How to use and add DataLoaders in the Commons backend. Use when implementing batched lookups, adding a new loader, wiring field resolvers to loaders, or avoiding N+1 via request-scoped batching and caching.
Use the GitHub CLI (gh) for PRs, repos, search, releases, and API access from the terminal. Use when attempting to get information about other repos or about PRs in this repo.
Use Graphite CLI (gt) for stacked PRs and branch management. Use when creating branches, committing changes, submitting PRs, syncing with trunk, or managing stacked pull requests.