Dev Tools
An agent that searches coding and API docs
This example shows how the Nimble task agent API answers software library and API usage questions by searching official documentation, changelogs, and release notes for the current answer, then returns a working code snippet and the source URL used.
Quick Start
Inputs
- Library or API question A natural-language question about how to use a library, framework, or API, typed at the command line.
Outputs
- A structured answer explaining the current recommended usage, including any breaking changes or migration notes
- A complete, working code snippet demonstrating the current API
- One or more citation URLs pointing to the official documentation or changelog used
- A local history log of every past question and answer, browsable and editable from the command line
How it works
A five-step pipeline built on the Task Agents API.
- Create the agent A Nimble Web Search Agent is configured once through the Task Agents API with domain expertise, goals, an allow list of official documentation and changelog domains, and a structured output schema.
- Run and search Each question starts a run, and the agent searches only the allowed documentation, GitHub release notes, and changelog sources for the current answer instead of relying on the model's training data.
- Poll to completion The client polls the run's status until it finishes, showing a live progress bar benchmarked against the agent's effort tier.
- Return structured output Nimble returns the parsed result matching the output schema: library, version, answer, code snippet, and citation URLs.
- Log and browse history Every completed answer is appended to a local history file that can be browsed, viewed, or deleted from the same command-line session.
Stack
Nimble primitives plus the full runtime stack.
Nimble APIs
What it does
- Task Agent API A custom-configured Task Agents API agent scoped to official docs, GitHub release notes, and changelogs; every run returns cited, structured JSON matching a defined output schema.
3rd Party Tools
Role
- Python Runs the command-line agent: configuration loading, the HTTP client, terminal rendering, and history storage.
- httpx Makes the REST calls to the Task Agents API to create the agent, start runs, poll status, and fetch results.
Reach out if you have any questions.
Talk to an Expert