# Send Fastapi Studio Template to your agent Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually. ## Fast path - Download the package from Yavira. - Extract it into a folder your agent can access. - Paste one of the prompts below and point your agent at the extracted folder. ## Suggested prompts ### New install ```text I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete. ``` ### Upgrade existing ```text I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "fastapi-studio-template", "name": "Fastapi Studio Template", "source": "tencent", "type": "skill", "category": "开发工具", "sourceUrl": "https://clawhub.ai/nissan/fastapi-studio-template", "canonicalUrl": "https://clawhub.ai/nissan/fastapi-studio-template", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/fastapi-studio-template", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=fastapi-studio-template", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "SKILL.md" ], "downloadMode": "redirect", "sourceHealth": { "source": "tencent", "slug": "fastapi-studio-template", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-05-03T04:17:38.526Z", "expiresAt": "2026-05-10T04:17:38.526Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=fastapi-studio-template", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=fastapi-studio-template", "contentDisposition": "attachment; filename=\"fastapi-studio-template-1.2.2.zip\"", "redirectLocation": null, "bodySnippet": null, "slug": "fastapi-studio-template" }, "scope": "item", "summary": "Item download looks usable.", "detail": "Yavira can redirect you to the upstream package for this item.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/fastapi-studio-template" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] } }, "links": { "detailUrl": "https://openagent3.xyz/skills/fastapi-studio-template", "downloadUrl": "https://openagent3.xyz/downloads/fastapi-studio-template", "agentUrl": "https://openagent3.xyz/skills/fastapi-studio-template/agent", "manifestUrl": "https://openagent3.xyz/skills/fastapi-studio-template/agent.json", "briefUrl": "https://openagent3.xyz/skills/fastapi-studio-template/agent.md" } } ``` ## Documentation ### FastAPI Studio Template Bootstrap a dark-themed FastAPI + HTMX studio app for generative AI comparison, A/B testing, and human evaluation with real-time progress streaming. ### When to Use Any "studio" app: image generation comparison, text model A/B testing, human evaluation UI Apps needing real-time progress updates (generation can take 30s–15min) Blind test / evaluation interfaces where raters shouldn't know which model produced which output Rapid prototyping of gen AI comparison tools ### When NOT to Use Simple CRUD apps (use standard FastAPI + Jinja2) Apps that don't need real-time progress (SSE adds complexity) Production-scale apps with 100+ concurrent users (use WebSockets instead of SSE) ### SSE Async Pattern (Critical) MUST use threading.SimpleQueue + asyncio polling. Do NOT use run_in_executor with blocking reads — it deadlocks the event loop. import asyncio import threading from queue import SimpleQueue from fastapi import FastAPI from fastapi.responses import StreamingResponse app = FastAPI() async def event_stream(queue: SimpleQueue): """Yield SSE events from a thread-safe queue.""" while True: try: msg = queue.get_nowait() except Exception: await asyncio.sleep(0.1) continue if msg is None: # sentinel yield f"data: {{\\"done\\": true}}\\n\\n" break yield f"data: {msg}\\n\\n" @app.get("/generate/stream") async def generate_stream(prompt: str, model: str): queue = SimpleQueue() def _run(): # Heavy generation work in background thread for step in range(10): import time; time.sleep(1) queue.put(f'{{"step": {step}, "total": 10}}') queue.put(None) # done sentinel threading.Thread(target=_run, daemon=True).start() return StreamingResponse(event_stream(queue), media_type="text/event-stream") Why not run_in_executor? FastAPI's executor runs on a thread pool, but SSE needs to yield events incrementally. Blocking in the executor means you can't stream partial progress — you'd have to wait for the entire generation to finish. The queue pattern decouples generation from streaming. ### Blind Test Mode Generate N variants (one per model), randomise display order, reveal model identity only after the user rates all variants. import random import uuid def create_blind_test(prompt: str, models: list[str]) -> dict: test_id = str(uuid.uuid4()) variants = [] for model in models: variants.append({ "variant_id": str(uuid.uuid4()), "model": model, # hidden from UI until reveal "prompt": prompt, }) random.shuffle(variants) return { "test_id": test_id, "variants": variants, "display_order": [v["variant_id"] for v in variants], } In the HTMX frontend, render variants as "Option A", "Option B", etc. On rating submission, return the mapping from option letters to model names. ### Hot-Loaded Model Singleton (ModelRegistry) Cold-loading SDXL or similar models takes 6–14 minutes. Cache loaded models in a registry singleton. class ModelRegistry: _instance = None _models: dict = {} _lock = threading.Lock() @classmethod def get(cls, model_name: str): with cls._lock: if model_name not in cls._models: cls._models[model_name] = cls._load_model(model_name) return cls._models[model_name] @classmethod def _load_model(cls, name: str): # Import and load the model if name == "sdxl": from mflux import Flux1 return Flux1.from_alias("schnell", quantize=8) raise ValueError(f"Unknown model: {name}") Preload at startup via the FastAPI lifespan hook for models you know you'll need. ### float32 Requirement for SDXL on MPS torch 2.10 on Apple Silicon (MPS) produces NaN outputs with float16 for SDXL. Force float32: import torch torch.set_default_dtype(torch.float32) # or per-model: model = model.to(dtype=torch.float32) This doubles VRAM usage but is the only reliable option until the MPS float16 bug is fixed. ### SQLite Schema for Ratings CREATE TABLE ratings ( id INTEGER PRIMARY KEY AUTOINCREMENT, test_id TEXT NOT NULL, variant_id TEXT NOT NULL, model TEXT NOT NULL, rater TEXT DEFAULT 'anonymous', score INTEGER CHECK(score BETWEEN 1 AND 5), preferred BOOLEAN DEFAULT FALSE, -- winner of pairwise comparison notes TEXT, created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ); CREATE INDEX idx_ratings_test ON ratings(test_id); CREATE INDEX idx_ratings_model ON ratings(model); ### Langfuse Tracing Wrap generation calls with Langfuse traces for cost tracking and latency monitoring: from langfuse import Langfuse langfuse = Langfuse() def generate_with_trace(prompt, model_name): trace = langfuse.trace(name="studio-generation", metadata={"model": model_name}) span = trace.span(name="generate", input={"prompt": prompt}) result = ModelRegistry.get(model_name).generate(prompt) span.end(output={"length": len(result)}) return result ### Worked Example: Minimal Studio App """Minimal FastAPI+HTMX studio with SSE progress.""" import asyncio import json import threading from queue import SimpleQueue from pathlib import Path from fastapi import FastAPI, Request from fastapi.responses import HTMLResponse, StreamingResponse from fastapi.staticfiles import StaticFiles app = FastAPI() HTML = """ Studio

🎨 Studio



""" @app.get("/", response_class=HTMLResponse) async def index(): return HTML async def event_stream(queue: SimpleQueue): while True: try: msg = queue.get_nowait() except Exception: await asyncio.sleep(0.1) continue if msg is None: yield f"data: {json.dumps({'done': True})}\\n\\n" break yield f"data: {msg}\\n\\n" @app.get("/generate/stream") async def generate_stream(prompt: str): queue = SimpleQueue() def _run(): import time for i in range(10): time.sleep(0.5) queue.put(json.dumps({"step": i + 1, "total": 10})) queue.put(None) threading.Thread(target=_run, daemon=True).start() return StreamingResponse(event_stream(queue), media_type="text/event-stream") Run with: uvicorn app:app --reload --port 8000 ### Tips Dark theme first — gen AI studios are used in long sessions; light themes cause eye strain Always show progress — users will close the tab if they think it's frozen Log every generation — Langfuse traces are invaluable for debugging quality issues Rate-limit generation — SDXL on MPS can only do one image at a time; queue requests Export ratings as CSV — researchers need data in portable formats ## Trust - Source: tencent - Verification: Indexed source record - Publisher: nissan - Version: 1.2.1 ## Source health - Status: healthy - Item download looks usable. - Yavira can redirect you to the upstream package for this item. - Health scope: item - Reason: direct_download_ok - Checked at: 2026-05-03T04:17:38.526Z - Expires at: 2026-05-10T04:17:38.526Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/fastapi-studio-template) - [Send to Agent page](https://openagent3.xyz/skills/fastapi-studio-template/agent) - [JSON manifest](https://openagent3.xyz/skills/fastapi-studio-template/agent.json) - [Markdown brief](https://openagent3.xyz/skills/fastapi-studio-template/agent.md) - [Download page](https://openagent3.xyz/downloads/fastapi-studio-template)