NanoAgent

An AI coding agent for desktop and terminal workflows.

NanoAgent brings an AI teammate into your everyday engineering flow. Open a project, ask for help, and let the agent inspect code, explain behavior, make focused changes, run validation commands, and review work with clear permission boundaries.

It is designed for developers who want useful automation without giving up control. NanoAgent can move through real tasks, but sensitive actions such as edits, command execution, network access, memory writes, MCP tools, and elevated operations remain governed by profiles, permissions, and approval prompts.

Why NanoAgent

Work from a desktop app or the nanoai terminal command.
Ask for feature work, bug fixes, planning, code review, and build/test loops.
Run NanoAI review automation from GitHub, GitLab, or Bitbucket PR/MR workflows.
Choose OpenAI, OpenAI ChatGPT Plus/Pro sign-in, OpenRouter, Anthropic, Google AI Studio, or an OpenAI-compatible provider.
Switch between hands-on build mode, read-only planning, and read-only review.
Delegate focused work to built-in or project-defined subagents.
Add project instructions, skills, custom agents, process-based custom tools, MCP tools, and reusable lessons.
Keep risky actions visible with permission prompts, policy rules, and undo/redo for tracked file edits.

Product Experience

Desktop

Use the desktop app when you want a visual workspace with sections, model controls, slash-command suggestions, permission prompts, activity output, and undo/redo close at hand.

Terminal

Use nanoai when you want a keyboard-first workflow, one-shot prompts, piped input, or quick review and automation from the command line.

nanoai
nanoai "Summarize this repository"
echo "Review the latest changes for regressions" | nanoai --profile review

CI Review Automation

The included GitHub Actions, GitLab CI, and Bitbucket Pipelines examples install NanoAI from the latest release, run the workspace pr-reviewer profile against the PR/MR diff, and post a review comment.

Copy .nanoagent/agents/pr-reviewer.md plus the matching CI files for your platform: .github/workflows/nanoai-review.yml and .github/nanoai-github-review.sh, .gitlab-ci.yml and .gitlab/nanoai-gitlab-review.sh, or bitbucket-pipelines.yml and .bitbucket/nanoai-bitbucket-review.sh.

Configure NANOAGENT_API_KEY. GitLab posting needs GITLAB_TOKEN or NANOAI_GITLAB_TOKEN; Bitbucket posting needs BITBUCKET_ACCESS_TOKEN or BITBUCKET_USERNAME plus BITBUCKET_APP_PASSWORD. Optional variables are NANOAGENT_PROVIDER, NANOAGENT_MODEL, NANOAGENT_BASE_URL, and NANOAGENT_THINKING.

Profiles

Profile	Best for
`build`	Implementation, fixes, tests, and validation.
`plan`	Read-only investigation and implementation plans.
`review`	Read-only code review focused on bugs, regressions, and missing tests.
`general`	Bounded delegated implementation work.
`explore`	Fast read-only project investigation.

Providers

Provider	Setup
OpenAI	API key
OpenAI ChatGPT Plus/Pro	Browser sign-in
OpenRouter	API key
Google AI Studio	API key
Anthropic	API key
OpenAI-compatible provider	Base URL and API key

Install

Desktop Downloads

Platform	Download
Windows x64	Installer
Linux x64	Zip
Linux arm64	Zip
macOS x64	Zip
macOS arm64	Zip

CLI

macOS / Linux:

curl -fsSL https://raw.githubusercontent.com/rizwan3d/NanoAgent/master/scripts/install.sh | bash

Windows PowerShell:

irm https://raw.githubusercontent.com/rizwan3d/NanoAgent/master/scripts/install.ps1 | iex

Restart your shell if nanoai is not immediately available.

First Run

Start NanoAgent:

nanoai

NanoAgent will guide you through provider setup, model discovery, and the first section. After setup, you can switch models with the terminal F2 or /models picker, or switch profiles and thinking mode from the desktop controls or terminal commands.

For terminal onboarding, you can pass an API key up front:

nanoai --provider-auth-key <key>

Common Commands

Command	Purpose
`/help`	Show available commands.
`/config`	Show provider, model, section, profile, thinking mode, and config path.
`/models`	Choose the active model with the arrow-key picker.
`/use <model>`	Switch directly to a model id.
`/onboard`	Re-run provider onboarding and switch the active session to the new provider.
`/profile <name>`	Switch profile.
`/thinking [on\|off]`	Show or set thinking mode.
`/permissions`	Show permission policy summary.
`/rules`	Show effective rules.
`/allow <tool-or-tag> [pattern]`	Add a session allow override.
`/deny <tool-or-tag> [pattern]`	Add a session deny override.
`/mcp`	Show MCP servers, custom tool providers, and dynamic tools.
`/init`	Create `.nanoagent` starter files for a project.
`/update [now]`	Check for updates, or install immediately with `/update now`.
`/undo`	Roll back the most recent tracked file edit transaction.
`/redo`	Re-apply the most recently undone edit transaction.
`/exit`	Exit the terminal UI.

Press F2 in the terminal UI to choose the active model with the same arrow-key picker.
Type / in the terminal input to open command suggestions, then use Up/Down and Enter to choose a command.

Safety and Control

NanoAgent is built around explicit control:

build, plan, and review profiles shape what the agent is allowed to do.
Permission rules decide whether actions are allowed, denied, or require approval.
Sensitive actions can prompt before they run.
Session overrides let you allow or deny a tool pattern temporarily.
Tracked file edits can be undone and redone.
Secret-looking values are redacted before logs, memory, audit records, and displayed tool output.

Your code stays on your machine. Prompts, relevant snippets, tool output, and conversation context are sent to the model provider you choose when they are needed for a request.

Learn More

The detailed user guide lives in docs/documentation.md. It covers onboarding, desktop and terminal workflows, providers, models, permissions, MCP, memory, hooks, custom agents, troubleshooting, and source builds.

License

Apache License 2.0. See LICENSE.