The Qodo Gen CLI lets you interact with the Qodo platform from your terminal for automation, advanced AI workflows, or CI/CD integration.
Use tools like Qodo Gen, Qodo Merge and Qodo Aware directly via the command line.
- Run interactive AI agents from your terminal
- Choose specific AI models on the fly
- Integrate with your own tools and schemas
- Serve agents over HTTP with
--mcpmode
To use Qodo CLI, you’ll need Node.js and npm installed.
Then run:
npm install -g @qodo/genqodo loginThis will open a browser for you to authenticate.
After logging in, you'll receive an API key to use locally or in CI. Your API key will be saved in the .qodo folder in your home dir, and also displayed for your reference.
qodo chatYou can use all Qodo tools in natural language:
Use Qodo Merge to summarize my latest git changes.
Exit anytime with Escape.
Create an agent.toml:
qodo initThe agent.toml file is used to configure your own agent. Learn more on how to do this in Qodo’s documentation platform.
For example:
description = "" # a description of what your agent does
instructions = """ """ # a prompt for the AI models explailing the required behavior
arguments = [{}, {}...] # a list of possible arguments that can be given to the agent
mcpServers = """ """ # list of MCP servers used by the agent
available_tools = [] # list of MCP server names
execution_strategy = "act" # plan lets the agent think through a multi-step strategy, act executes actions immediately
output_schema = """ {} """ # valid json of the wanted agent output
exit_expression = "include_tests" # for CI runs, a condition used to determine if the agent run succeeded or failed
Run your agent with:
qodo my-commandWhen you run an agent with the --mcp flag, it starts a local HTTP server on port 3000. This turns the agent into a standalone AI tool that can receive and respond to requests over HTTP:
qodo my-command --mcpqodo --help # See all commands
qodo login # Authenticate
qodo chat # Start interactive agent session (Qodo Gen)
qodo init # Generate a customizable agent template
qodo <command> # Run your configured agent
qodo mcp-list # List all available tools
qodo models # List supported AI modelsmcp.json: lists all the shared tools across agentsagent.toml: defines an agent, with tools and output schemas
You can copy the API key shown after login into your CI/CD secrets. The key is tied to your user and subject to the same limits.
This repository includes a GitHub Action that makes it easy to integrate Qodo into your GitHub workflows.
- uses: qodo-ai/qodo-gen-cli@main
with:
prompt: "review"By default, this uses the example agent.toml included with the action. To use your own agent configuration, specify the agentfile input:
- uses: qodo-ai/qodo-gen-cli@main
with:
prompt: "review"
agentfile: "./my-custom-agent.toml" # Path relative to your repository root- uses: qodo-ai/qodo-gen-cli@main
with:
prompt: "your prompt here" # Required: The prompt or command to run
model: "claude-4-sonnet" # Optional: Specify AI model
agentfile: "path/to/agent.toml" # Optional: Custom agent file (default: examples/agent.toml from action)
qodo-version: "latest" # Optional: @qodo/gen version (default: latest)
key-value-pairs: | # Optional: Additional parameters
coverage_threshold=80
test_framework=jestYou can pass additional parameters in two formats:
JSON Format:
key-value-pairs: |
{
"coverage_score_threshold": "0.8",
"max_issues": "10",
"include_suggestions": "true"
}Multiline Format:
key-value-pairs: |
test_framework=jest
coverage_threshold=90
parallel=trueExample: Automated Test Coverage Generation with Label Trigger
This example shows how to create a test coverage bot that automatically generates tests for uncovered code when a specific label is added to a PR:
name: Test Coverage Bot
on:
pull_request:
branches:
- main
types:
- labeled
permissions:
pull-requests: write
contents: write
jobs:
coverage:
# Only run if:
# 1. PR has the qodo-cover label
# 2. PR is open (not closed or draft)
if: |
contains(github.event.label.name, 'qodo-cover') &&
github.event.pull_request.state == 'open' &&
github.event.pull_request.draft == false
runs-on: ubuntu-latest
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Run Qodo Coverage Bot
uses: qodo-ai/qodo-gen-cli@v1
with:
prompt: "qodo-cover"
# agentfile: "${{ github.workspace }}/agent.toml"
key-value-pairs: |
desired_coverage=90
env:
QODO_API_KEY: ${{ secrets.QODO_API_KEY }}
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
PR_NUMBER: ${{ github.event.pull_request.number }}In this example:
- The workflow triggers when a label is added to a PR
- It only runs if the label is "qodo-cover" and the PR is open
- It uses the
qodo-coverprompt to:- Analyze which changed files need test coverage
- Generate appropriate tests for uncovered code
- Create a follow-up PR with the new tests targeting the original PR branch
- Sets a desired coverage threshold of 90%
- Requires both
QODO_API_KEYandGITHUB_TOKENsecrets - Passes the PR number as an environment variable for the bot to reference
Note:
- Make sure to set your
QODO_API_KEYin your repository secrets. - allow GitHub Actions to create pull requests. This setting can be found under: Settings > Actions > General > Workflow permissions (near the bottom of the page).
See ./examples/agents/cover.toml for the agent configuration.
Example: GitHub Comment Mention Bot
name: Qodo Mention Bot
on:
issue_comment:
types: [created]
pull_request_review_comment:
types: [created]
permissions:
issues: write
pull-requests: write
contents: write
jobs:
respond:
if: ${{ startsWith(github.event.comment.body, '/qodo ') && github.event.comment.user.login == github.repository_owner }}
runs-on: ubuntu-latest
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Run Qodo Mention Bot
uses: qodo-ai/qodo-gen-cli@v1
with:
prompt: "qodo-mention"
# agentfile: "${{ github.workspace }}/agent.toml"
key-value-pairs: |
event_path=${{ github.event_path }}
env:
QODO_API_KEY: ${{ secrets.QODO_API_KEY }}
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
This example shows how to create a bot that responds to comments mentioning "/qodo" in a GitHub issue or PR comment. For example: "/qodo can you explain this issue?"
See ./examples/agents/mention.toml for the agent configuration.
Example: Release Notes Bot
name: Release Notes Generator (manual)
on:
workflow_dispatch:
inputs:
target_tag:
description: "Generate notes up to (and including) this tag; blank = HEAD"
required: false
permissions:
contents: write
pull-requests: write
issues: read
id-token: write
jobs:
release_notes:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- name: Qodo release-notes agent
uses: qodo-ai/qodo-gen-cli@v1
with:
prompt: qodo-release-notes
# agentfile: "${{ github.workspace }}/agent.toml"
key-value-pairs: |
target_tag=${{ github.event.inputs.target_tag }}
notes_file=RELEASE_NOTES.md
env:
QODO_API_KEY: ${{ secrets.QODO_API_KEY }}
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
This example shows how to create a manually-dispatched bot that generates release notes for a given tag. The bot will create a new PR that updates the release notes file with the new release notes.
See ./examples/agents/release.toml for the agent configuration.
For full documentation, visit the Qodo documentation website.