SHIPIT

AdaptiveAgent

1 min read
6 sections
Edit this page

Agent that creates new tools at runtime from Python code. When it needs a capability it doesn't have, it writes one.

Quick start

python
from shipit_agent.deep import AdaptiveAgent
from shipit_agent.tools.base import ToolContext

agent = AdaptiveAgent(llm=llm, can_create_tools=True)

fib = agent.create_tool(
    name="fibonacci",
    description="Calculate Nth Fibonacci number",
    code="""
    def fibonacci(n: int) -> str:
        a, b = 0, 1
        for _ in range(n):
            a, b = b, a + b
        return str(a)
    """,
)

# Test it
print(fib.run(ToolContext(prompt="test"), n=10).text)  # "55"

# Use in agent runs
result = agent.run("What is fibonacci of 20?")

With built-in tools

python
agent = AdaptiveAgent.with_builtins(llm=llm, can_create_tools=True)
# Has web search + code exec + dynamic tool creation

With Super RAG

python
from shipit_agent.rag import RAG, HashingEmbedder

rag = RAG.default(embedder=HashingEmbedder(dimension=512))
rag.index_file("docs/api.md")

agent = AdaptiveAgent.with_builtins(llm=llm, rag=rag)
# rag_search/rag_fetch_chunk are auto-wired alongside the runtime tool
# factory; the agent can write a new tool that calls into RAG.

Streaming

python
for event in agent.stream("Parse the CSV at /data/sales.csv and report totals"):
    print(f"[{event.type}] {event.message}")

Inside a DeepAgent

python
from shipit_agent.deep import DeepAgent

builder = AdaptiveAgent.with_builtins(llm=llm, name="builder")
deep = DeepAgent.with_builtins(llm=llm, agents=[builder])
deep.run("Use the builder sub-agent to create a CSV totals tool, then use it.")

Properties

PropertyTypeDescription
created_toolslist[CreatedTool]All dynamically created tools
toolslist[Tool]All tools (including created)

Note

Code strings are auto-dedented, so indented code in notebooks works fine.