Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Integrate with Replicate AI for running models (image generation, LLMs, etc.).
.claude/skills/aiskillstore-replicate-handler/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-21 | ✓→✗ | ▼ Worse | -10% | 0% |
npm install replicateREPLICATE_API_TOKEN to .env (and .env.local).For models that complete quickly (seconds), use replicate.run.
typescriptimport Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, }); // Run a model const output = await replicate.run( "owner/model:version", { input: { prompt: "..." } } );
For tasks that might timeout (video generation, large models), use Inngest's step.sleep to poll for completion.
typescript// In an Inngest function export const generateVideo = inngest.createFunction( { id: "generate-video" }, { event: "video.generate" }, async ({ event, step }) => { // 1. Create Prediction const prediction = await step.run("create-prediction", async () => { return await replicate.predictions.create({ version: "model-version-hash", input: { prompt: event.data.prompt } }); }); let status = prediction.status; let result = prediction; // 2. Poll for Completion while (status !== "succeeded" && status !== "failed" && status !== "canceled") { // Sleep for 5s (Inngest handles this without consuming server time) await step.sleep("wait-for-model", "5s"); // Check status result = await step.run("check-status", async () => { return await replicate.predictions.get(prediction.id); }); status = result.status; } // 3. Handle Result if (status === "failed") { throw new Error(`Replicate failed: ${result.error}`); } return result.output; } );
See reference.md for the full type definitions of the Replicate SDK.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 17,971 | 7,888 | -56% | 1 | 1 | 0% | 3,159 | 1,993 | -37% | 0 | 0 | — |
case-02 | fail→pass | 15,547 | 8,897 | -43% | 1 | 1 | 0% | 2,808 | 2,206 | -21% | 0 | 0 | — |
case-03 | pass→pass | 3,584 | 3,034 | -15% | 1 | 1 | 0% | 616 | 1,048 | +70% | 0 | 0 | — |
case-04 | pass→pass | 8,752 | 4,364 | -50% | 1 | 1 | 0% | 1,452 | 1,205 | -17% | 0 | 0 | — |
case-05 | fail→pass | 11,351 | 8,209 | -28% | 1 | 1 | 0% | 1,993 | 2,153 | +8% | 0 | 0 | — |
case-06 | pass→pass | 8,039 | 3,458 | -57% | 1 | 1 | 0% | 1,429 | 1,143 | -20% | 0 | 0 | — |
case-07 | pass→pass | 11,435 | 5,368 | -53% | 1 | 1 | 0% | 1,964 | 1,412 | -28% | 0 | 0 | — |
case-08 | pass→pass | 5,439 | 2,983 | -45% | 1 | 1 | 0% | 937 | 1,035 | +10% | 0 | 0 | — |
case-09 | pass→pass | 14,560 | 8,670 | -40% | 1 | 1 | 0% | 2,498 | 1,980 | -21% | 0 | 0 | — |
case-10 | pass→pass | 8,023 | 1,970 | -75% | 1 | 1 | 0% | 1,369 | 857 | -37% | 0 | 0 | — |
case-11 | pass→pass | 15,369 | 8,212 | -47% | 1 | 1 | 0% | 2,569 | 2,045 | -20% | 0 | 0 | — |
case-12 | pass→pass | 9,588 | 2,292 | -76% | 1 | 1 | 0% | 1,579 | 923 | -42% | 0 | 0 | — |
case-13 | pass→pass | 13,497 | 5,694 | -58% | 1 | 1 | 0% | 2,285 | 1,526 | -33% | 0 | 0 | — |
case-14 | pass→pass | 14,519 | 7,424 | -49% | 1 | 1 | 0% | 2,602 | 1,841 | -29% | 0 | 0 | — |
case-15 | pass→pass | 7,440 | 3,364 | -55% | 1 | 1 | 0% | 1,241 | 1,083 | -13% | 0 | 0 | — |
case-16 | fail→pass | 12,749 | 1,780 | -86% | 1 | 1 | 0% | 2,263 | 808 | -64% | 0 | 0 | — |
case-17 | pass→pass | 10,142 | 8,612 | -15% | 1 | 1 | 0% | 1,862 | 2,039 | +10% | 0 | 0 | — |
case-18 | pass→pass | 13,529 | 6,800 | -50% | 1 | 1 | 0% | 2,489 | 1,859 | -25% | 0 | 0 | — |
case-19 | pass→pass | 10,429 | 4,923 | -53% | 1 | 1 | 0% | 1,731 | 1,416 | -18% | 0 | 0 | — |
case-20 | pass→pass | 6,728 | 3,828 | -43% | 1 | 1 | 0% | 1,235 | 1,273 | +3% | 0 | 0 | — |
case-21 | pass→fail | 12,327 | 8,749 | -29% | 1 | 1 | 0% | 2,404 | 2,174 | -10% | 0 | 0 | — |
case-22 | fail→fail | 12,422 | 12,298 | -1% | 1 | 1 | 0% | 2,397 | 2,963 | +24% | 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. 22 cases were attempted. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.