The Ultimate AI Agent & Developer Automation Kit

Prompts, CLIs & Boilerplates — by Limit X

Ship AI-powered tools faster. This kit collects the exact prompts, command-line workflows and starter boilerplates used to build local-AI automations, agent pipelines and developer tooling — all open-source, all runnable on your own machine.

---

Part 1 — AI Agent Prompts (Copy-Paste)

1.1 The System-Prompt Skeleton (for any agent)

You are <ROLE>. Your job is to <OUTCOME>.
Rules:
- Work step by step. State your plan before acting.
- Use only the tools listed. Never invent a command, file path or API.
- If a step fails, report the exact error and stop — do not guess a fix.
- Finish with a 3-line summary: what changed, what was verified, what remains.

1.2 Code-Review Agent Prompt

Review the code below like a senior engineer.
Output format:
1. BUGS — anything that will break in production (file:line + why).
2. SECURITY — injection, secrets, auth gaps.
3. PERF — N+1 queries, unbounded loops, missing indexes.
4. STYLE — only issues that affect readability.
Be specific. No generic advice. If the code is fine, say so in one line.

1.3 Research Agent Prompt

Research <TOPIC>. Use primary sources only (official docs, repos, papers).
For every claim, give: the claim, the source URL, and the date you verified it.
Separate FACTS (sourced) from INFERENCE (your reasoning). End with an
"Unverified" section listing anything you could not confirm.

1.4 Debugging Agent Prompt

Debug this failure. Error: <PASTE ERROR>. Code: <PASTE CODE>.
Method:
1. State the 3 most likely root causes, ranked.
2. For each, give ONE test that confirms or rules it out.
3. Recommend the smallest fix first. Never rewrite the whole file.

---

Part 2 — CLI Workflows

2.1 Local LLM with Ollama

# Install + pull a model + run
curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.1:8b
ollama run llama3.1:8b

# Serve an API on localhost:11434
ollama serve
curl http://localhost:11434/api/generate -d '{"model":"llama3.1:8b","prompt":"Why is the sky blue?"}'

2.2 Custom Modelfile (your own tuned model)

FROM llama3.1:8b
PARAMETER temperature 0.3
SYSTEM """
You are a concise coding assistant. Answer with code first, explanation after.
"""

ollama create coder -f Modelfile
ollama run coder

2.3 Web Scraping with crawl4ai

pip install crawl4ai
crawl4ai-setup

import asyncio
from crawl4ai import AsyncWebCrawler

async def main():
    async with AsyncWebCrawler() as crawler:
        result = await crawler.arun(url="https://example.com")
        print(result.markdown)

asyncio.run(main())

2.4 Serve a Model with vLLM

pip install vllm
vllm serve meta-llama/Llama-3.1-8B-Instruct --port 8000
# OpenAI-compatible API at http://localhost:8000/v1

2.5 Download Open Weights from Hugging Face

pip install huggingface_hub
hf download <org>/<model-repo>

---

Part 3 — Boilerplates

3.1 Minimal Agent Loop (Python)

import subprocess, json

def run_tool(cmd: list[str]) -> str:
    """Run a shell tool and return its output (trimmed)."""
    out = subprocess.run(cmd, capture_output=True, text=True, timeout=60)
    return (out.stdout + out.stderr)[:4000]

def agent_step(plan: str) -> dict:
    """One agent step: plan -> act -> observe. Extend with your LLM call."""
    return {"plan": plan, "status": "ok"}

if __name__ == "__main__":
    print(run_tool(["echo", "agent ready"]))

3.2 RAG in ~40 Lines (local embeddings)

from sentence_transformers import SentenceTransformer
import numpy as np

model = SentenceTransformer("all-MiniLM-L6-v2")
docs = ["Your document text here.", "Another chunk."]
emb = model.encode(docs)

def search(query: str, k: int = 2):
    q = model.encode([query])
    scores = (emb @ q.T).ravel()
    return [docs[i] for i in np.argsort(-scores)[:k]]

print(search("your question"))

3.3 FastAPI Wrapper for Any Local Model

from fastapi import FastAPI
import requests

app = FastAPI()

@app.post("/ask")
def ask(prompt: str):
    r = requests.post("http://localhost:11434/api/generate",
                      json={"model": "llama3.1:8b", "prompt": prompt, "stream": False})
    return {"answer": r.json()["response"]}

---

Part 4 — Automation Checklists

**Daily content pipeline checklist**

**Local-AI workstation checklist**

---

Get More

New tools, commands and deep-dives land daily:

**Newsletter:** https://hebis-newsletter.beehiiv.com/

**Daily hub:** https://limitx-hub.vercel.app

*Limit X — Open Source AI. Built in the open, for builders.*