Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Configure Hex local development with hot reload and testing. Use when setting up a development environment, configuring test workflows, or establishing a fast iteration cycle with Hex. Trigger with phrases like "hex dev setup", "hex local development", "hex dev environment", "develop with hex".
.claude/skills/jeremylongshore-hex-local-dev-loop/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 160% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -23% | 0% |
Set up a development workflow for Hex API orchestration with mocked API responses and testing.
hex-orchestrator/
├── src/hex/
│ ├── client.ts # API client
│ ├── orchestrator.ts # Pipeline runner
│ └── types.ts # TypeScript interfaces
├── tests/
│ ├── fixtures/ # Mock API responses
│ └── orchestrator.test.ts
├── .env.local
└── package.jsontypescript// src/hex/client.ts export class HexClient { constructor(private token: string, private baseUrl = 'https://app.hex.tech/api/v1') {} async listProjects() { return this.get('/projects'); } async runProject(projectId: string, inputParams?: Record<string, any>) { return this.post(`/project/${projectId}/run`, { inputParams: inputParams || {}, updateCacheResult: true }); } async getRunStatus(projectId: string, runId: string) { return this.get(`/project/${projectId}/run/${runId}`); } async cancelRun(projectId: string, runId: string) { return this.delete(`/project/${projectId}/run/${runId}`); } private async get(path: string) { const res = await fetch(`${this.baseUrl}${path}`, { headers: { 'Authorization': `Bearer ${this.token}` } }); if (!res.ok) throw new Error(`Hex API ${res.status}`); return res.json(); } private async post(path: string, body: any) { const res = await fetch(`${this.baseUrl}${path}`, { method: 'POST', headers: { 'Authorization': `Bearer ${this.token}`, 'Content-Type': 'application/json' }, body: JSON.stringify(body) }); if (!res.ok) throw new Error(`Hex API ${res.status}`); return res.json(); } private async delete(path: string) { const res = await fetch(`${this.baseUrl}${path}`, { method: 'DELETE', headers: { 'Authorization': `Bearer ${this.token}` } }); return res.ok; } }
typescriptimport { describe, it, expect, vi } from 'vitest'; const mockFetch = vi.fn(); vi.stubGlobal('fetch', mockFetch); describe('HexClient', () => { it('should list projects', async () => { mockFetch.mockResolvedValueOnce({ ok: true, json: async () => [{ projectId: 'p1', name: 'Test' }] }); const client = new HexClient('test-token'); const projects = await client.listProjects(); expect(projects).toHaveLength(1); }); it('should trigger a run', async () => { mockFetch.mockResolvedValueOnce({ ok: true, json: async () => ({ runId: 'r1', projectId: 'p1' }) }); const client = new HexClient('test-token'); const run = await client.runProject('p1', { date: '2025-01-01' }); expect(run.runId).toBe('r1'); }); });
See hex-sdk-patterns for production patterns.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-18 | pass→pass | 11,862 | 7,467 | -37% | 1 | 1 | 0% | 1,043 | 1,238 | +19% | 0 | 0 | — |
case-01 | fail→pass | 19,512 | 22,685 | +16% | 1 | 1 | 0% | 4,195 | 4,650 | +11% | 0 | 0 | — |
case-02 | pass→pass | 18,369 | 21,395 | +16% | 1 | 1 | 0% | 3,737 | 4,512 | +21% | 0 | 0 | — |
case-03 | fail→pass | 28,953 | 27,961 | -3% | 1 | 1 | 0% | 5,554 | 5,977 | +8% | 0 | 0 | — |
case-04 | pass→pass | 27,941 | 27,984 | +0% | 1 | 1 | 0% | 3,944 | 5,629 | +43% | 0 | 0 | — |
case-05 | pass→pass | 15,469 | 20,705 | +34% | 1 | 1 | 0% | 2,264 | 3,722 | +64% | 0 | 0 | — |
case-06 | pass→pass | 21,357 | 11,871 | -44% | 1 | 1 | 0% | 3,003 | 3,222 | +7% | 0 | 0 | — |
case-07 | pass→pass | 14,648 | 3,709 | -75% | 1 | 1 | 0% | 1,802 | 1,567 | -13% | 0 | 0 | — |
case-08 | fail→fail | 15,453 | 18,540 | +20% | 1 | 1 | 0% | 1,806 | 3,627 | +101% | 0 | 0 | — |
case-09 | fail→pass | 7,142 | 3,800 | -47% | 1 | 1 | 0% | 1,234 | 1,594 | +29% | 0 | 0 | — |
case-10 | fail→pass | 8,693 | 8,983 | +3% | 1 | 1 | 0% | 625 | 1,624 | +160% | 0 | 0 | — |
case-11 | fail→pass | 25,447 | 11,465 | -55% | 1 | 1 | 0% | 2,665 | 2,046 | -23% | 0 | 0 | — |
case-12 | fail→fail | 18,281 | 17,608 | -4% | 1 | 1 | 0% | 2,057 | 3,061 | +49% | 0 | 0 | — |
case-13 | fail→fail | 23,329 | 19,518 | -16% | 1 | 1 | 0% | 2,799 | 3,835 | +37% | 0 | 0 | — |
case-14 | pass→pass | 16,646 | 14,970 | -10% | 1 | 1 | 0% | 2,116 | 2,244 | +6% | 0 | 0 | — |
case-15 | fail→pass | 11,522 | 12,537 | +9% | 1 | 1 | 0% | 2,113 | 2,215 | +5% | 0 | 0 | — |
case-16 | pass→pass | 12,942 | 1,895 | -85% | 1 | 1 | 0% | 1,359 | 1,123 | -17% | 0 | 0 | — |
case-17 | fail→pass | 15,347 | 6,865 | -55% | 1 | 1 | 0% | 1,655 | 1,150 | -31% | 0 | 0 | — |
case-19 | pass→pass | 10,218 | 3,034 | -70% | 1 | 1 | 0% | 1,770 | 1,423 | -20% | 0 | 0 | — |
case-20 | fail→pass | 16,946 | 8,156 | -52% | 1 | 1 | 0% | 2,007 | 1,211 | -40% | 0 | 0 | — |
case-21 | pass→pass | 16,110 | 14,497 | -10% | 1 | 1 | 0% | 1,870 | 2,178 | +16% | 0 | 0 | — |
case-22 | pass→pass | 22,621 | 8,907 | -61% | 1 | 1 | 0% | 2,558 | 1,445 | -44% | 0 | 0 | — |
case-23 | pass→pass | 14,327 | 9,433 | -34% | 1 | 1 | 0% | 1,714 | 1,671 | -3% | 0 | 0 | — |
case-24 | pass→pass | 14,262 | 14,149 | -1% | 1 | 1 | 0% | 2,702 | 2,518 | -7% | 0 | 0 | — |
case-25 | pass→pass | 30,090 | 7,727 | -74% | 1 | 1 | 0% | 1,785 | 1,303 | -27% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 25 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.