What the Agent can help with
The Agent can help with work across the Codemod platform:- Campaigns: review campaign progress, inspect failed or stuck tasks, and open the campaign creation form for a registry package.
- Automations: explain triggers and actions, identify missing guardrails, and help configure automation steps.
- Insights: inspect dashboards, create widgets, and reason about migration readiness or adoption trends.
- Grep: search indexed repositories with Pattern, Regex, or YAML queries, and create Insights dashboards from a completed search.
- Registry packages: search for relevant codemods and inspect package details before applying them.
- Repositories: use indexed repository context to search files, dependencies, and code patterns when repository access is available.
- Codemod creation: turn repeatable transformations or analysis ideas into generated codemod packages.
The Agent uses the current page as context. If you ask about a Campaign, Automation, or Insights dashboard while viewing it, the Agent can use that page state to give more specific help.
Start a conversation
1
Open the Agent
Select Agent in the sidebar for a full-page conversation, or press
Cmd+J on macOS or Ctrl+J on Windows and Linux to open it in a side panel from any page.2
Ask from the current page
Ask about the Campaign, Automation, Insights dashboard, Registry package, or repository context you are viewing.
3
Review suggested actions
When the Agent needs to navigate, create follow-up work, or choose between next steps, review the suggested action before continuing.
Grep answers and Insights dashboards
When you ask a question about your code, such as where an API is called or how often a pattern appears, the Agent answers in the conversation with a Grep search. Grep results reflect the current index and return quickly. Ask for an Insights dashboard when you want results that stay up to date: trends over time, migration or ownership tracking, or a view to share with your team. Building a dashboard takes longer, but its widgets keep refreshing as your code changes. For example:
Build an Insights dashboard that tracks fetchUser call sites by repository.
Questions about which repositories depend on a package use the Insights dependency inventory, because it covers dependencies across all indexed repositories.
Investigation mode (@investigate) has been retired. Ask your question about code behavior directly, and the Agent answers with Grep searches.
Build codemod packages
For repeatable code transformations or reusable analysis, the Agent can generate a codemod package and publish it to the registry. Generated packages are private by default unless you explicitly choose public visibility. When you are working in an organization, generated packages use that organization’s registry scope. After a package is ready, the Agent can continue with follow-up work such as creating a Campaign or building an Insights dashboard that uses the generated package. The same prompting guidance for building better codemods in Studio applies to the Agent: keep the transformation focused, describe the intended behavior clearly, and include examples when they help.Related docs
Campaigns
Run codemods with automated PRs and centralized tracking.
Automations
Create rules that react to events across your projects.
Insights
Create dashboards for migration and codebase visibility.
Grep
Search indexed private repositories with patterns, regex, or YAML rules.
Registry
Find and apply public, private, or pro codemod packages.
MCP
Let Claude, Cursor, and other AI assistants search your code and manage migrations.