Choose the SDK or CLI and start with a useful agent task

Build with your first agent

This page offers a clear overview of Build with your first agent. It helps you explore this topic and find a useful next step.

1: Create a Vidbyte API key (hosted CLI only)

For hosted research with vidbyte-cli, open API settings, create a key, and copy it somewhere secure immediately. Vidbyte only shows the plaintext key once. Skip this step if you are building a local agent with the SDK.

Create an API key

2: Fund hosted CLI research (optional)

Hosted research runs started with vidbyte-cli use API balance, funded separately from subscriptions through API settings. SDK agents use your configured model provider account instead, and local suggestion runs do not need Vidbyte API balance.

Add API balance

3: Choose an agent path

Build and run a Python agent with vidbyte-sdk, or use vidbyte-cli to start a hosted research run. SDK agents run with your model provider credentials; hosted CLI research uses the Vidbyte API key and balance from the optional setup above.

Try hosted research with the CLI

After installing vidbyte-cli and completing the optional API key and balance setup, start a research thread with a concrete question.

Start a research thread

vidbyte-cli research start "Compare three approaches to reducing latency in an LLM tool-calling workflow. Summarize tradeoffs and cite sources."

Build with the Vidbyte SDK

Create a Python agent, add tools and policies, then run it with your chosen model provider.

Read SDK documentation

Run with the Vidbyte CLI

Start hosted research threads or run supported local agent workflows from your terminal.

Read CLI documentation

4. Give your agent a concrete task

Choose the workflow that fits your problem: build a local Python agent with the SDK, start hosted research with the CLI, or run a local coding-agent ensemble. These examples are task prompts, not direct API requests.

Build a focused SDK agent

For example, ask an agent to compare two retry strategies for a service with strict latency and duplicate-write constraints. Install vidbyte-sdk and configure your model provider credentials first.

Run a Python agent

Python

from vidbyte import VidbyteSDK

agent = VidbyteSDK().agents.base(
    name="retry-reviewer",
    system_prompt="Compare engineering options against the stated constraints.",
    provider="openai",
    model_name="gpt-4.1",
)
reply = agent.run("Compare exponential backoff with jitter and fixed retries for an API client where duplicate writes are costly.")
print(reply.content)

Start hosted research

Use the CLI when you want Vidbyte-hosted research. The API key needs API balance; use the returned thread ID to continue with a follow-up question.

Research, then follow up

vidbyte-cli research start "Compare the main approaches to evaluating retrieval-augmented generation systems. Summarize tradeoffs and cite sources."
vidbyte-cli research add YOUR_THREAD_ID "Focus the comparison on evaluation cost and failure detection."

Review and improve a local codebase

The optional same-host ensemble runs through Codex on your machine. Install the Codex extra first; local runtime admission costs API balance, and model usage follows your provider subscription.

Explore code changes locally

python -m pip install "vidbyte-cli[codex]"
vidbyte-cli runtime same-host-ensemble "Review this repository's retry handling for duplicate side effects. Compare safe fixes, implement the strongest option, and run focused tests."

Turn a goal into next actions

Ask the local suggestion agent to produce ranked, actionable next steps and handoff details for a goal.

Generate next-step ideas

vidbyte-cli agents suggest run --goal "Turn our onboarding drop-off problem into five testable product experiments for the next two weeks." --count 5

SDK documentation

CLI command reference

skills.sh: code review

skills.sh: research reproduction