Skip to main content
Grep searches the latest indexed snapshots of repositories you can access in Codemod. Use it to find structural code patterns, text matches, or advanced ast-grep rules across your organization or personal Git account. Open Grep at app.codemod.com/grep.
Codemod Grep searches private and organization repositories you have granted Codemod access to. Public structural search across open-source repositories lives at grep.codemod.com.

Prerequisites

  • Sign in to Codemod.
  • Connect GitHub or GitLab so repositories appear in your scope.
  • Index the repositories you want to search. Listing and searching never start indexing automatically.
Repository scope follows the app context: the active organization when one is selected, or your personal Git repositories otherwise.

Search indexed code

1

Open Grep

Go to Grep in the Codemod app.
2

Index repositories (first run)

If no repositories are searchable yet, select the repositories to index and start indexing. Wait until they show as searchable before you run a query.When some repositories are already searchable, use Add repositories to index more without leaving the search studio.
3

Choose a mode and write a query

Pick Pattern, Regex, or YAML, then enter the query. Optionally narrow by language, repository, and path filter.
4

Review matches

Results group by repository and file, with snippets, exact line ranges, and links back to GitHub or GitLab. Use pagination to load more files when the result set continues.

Query modes

Prefer Pattern for source-code searches and YAML for multi-condition or constrained ast-grep rules. Use Regex when the target is genuinely textual, the language parser is unsupported, or a precise ast-grep query cannot be formed.
Pattern and YAML run through Codemod’s indexed ast-grep engine. Regex runs as a PostgreSQL regular expression against decrypted snapshot content inside the database. Interactive Grep does not execute user-authored JavaScript.

Filters

  • Language: search all indexed languages in scope, or limit to one or more parsers.
  • Repositories: omit the filter to search every searchable repository in the current scope, or select a subset.
  • Path: optional glob, contains, prefix, or basename filter to limit which files are considered.
YAML mode uses the language declared in the YAML document. That top-level language is authoritative for the rule.

Results

Successful searches return:
  • Match counts and truncated state when limits apply
  • File-grouped snippets with exact ranges (not full source files)
  • Provider links to the indexed commit on GitHub or GitLab
  • Partial failures for repositories that are not indexed, lack the selected language, or fail during execution

Check coverage before concluding that code is absent

Returned snippets remain useful even when some repositories fail. A zero-match page does not prove that code is absent when coverage is incomplete. If your search reports repository failures, inspect those failures and narrow the language, path, or expression within the scope you intended to search. Avoid repeatedly submitting the same failing query. When more pages are available, continue the same search. Changing the query, mode, repositories, language, or path starts a new search; do not reuse a continuation cursor from the old search. Reuse snippets and focused source ranges for files you already found.

Grep MCP

MCP clients can search the same repository scope through the Grep MCP endpoint:
Authenticate with a Codemod API key that has Use Grep enabled:
The server is read-only. It can list repositories, search indexed code, and read focused source ranges around matches. It does not start indexing or modify code. Index repositories in the Grep UI before searching from an MCP client. Store the key in a secret or environment variable. Do not commit it into MCP configuration files.

Select supported languages

For grep_search, use the scalar "all" or a non-empty array of parser names advertised in the tool’s input schema, such as ["go", "typescript"]. ["all"] is not a valid selection. Neither markdown nor makefile is a supported parser value. If a language selection is rejected, choose supported parsers that match your intended scope. Use "all" only when you intend to search all indexed languages; unsupported selections are not silently broadened.

Interpret search results

Check failures, countComplete, and nextCursor before making an absence or exhaustive-coverage claim:
  • Non-empty failures means some repository searches failed. Returned files are still usable evidence for their indexed commits.
  • countComplete: false means coverage is incomplete, even if nextCursor is null or no matches were returned. If there is no cursor, narrow your search before drawing an exhaustive conclusion.
  • A non-null nextCursor lets you continue with the identical search parameters. Start without the old cursor if you change the search.
The MCP response keeps its structured search payload and can include additional recovery guidance for incomplete results. A successful tool response does not mean that every repository search succeeded.

Create Insights dashboards from Grep

Use Grep for exploratory searches, then promote a useful query into a persistent Insights dashboard. The dashboard keeps the query and repository scope as editable configuration so you can track the same pattern over time—with Insights features such as timeseries trends, table breakdowns, and group-by dimensions like repository or codeowner (team ownership). Grep finds the matches. Insights turns that search into ongoing visibility.
1

Run a Grep search

Complete a Pattern, Regex, or YAML search against the repositories you care about. Refine language and path filters until the results look right.
2

Create an Insights dashboard

After results load, choose Create Insights dashboard. Name the dashboard, confirm which repositories to include (up to 10), and select the languages to track.
3

Review and extend in Insights

Open the new dashboard to see the generated match-total and repository-breakdown widgets. From there, add timeseries charts, ownership breakdowns, formulas, or additional queries using the normal Insights editor.
What Grep hands off:
  • The exact query, mode, optional path filter, and selected repositories
  • Count-oriented widgets for total matches and per-repository breakdown
  • Languages you chose at creation time (fewer languages usually compute faster)
What it does not copy into Insights:
  • Match snippets, file paths, or individual result rows from the Grep UI
After creation, the dashboard is a normal Insights dashboard: you can edit widgets, change repositories, and use template variables under the usual Insights permissions.
Prefer a focused language selection when creating the dashboard. Searching with language all in Grep is fine for discovery; for tracking, choose only the languages you need so Insights stays faster.
Regex dashboards use Insights/JSSG regex semantics, which can differ slightly from interactive Grep’s PostgreSQL regex results.

Troubleshooting

Insights

Track pattern adoption and migration progress with dashboards.

ast-grep YAML queries

Reuse ast-grep rules in Insights widgets.

Investigation

Turn Grep searches into a source-backed answer about code behavior.

GitHub integration

Grant repository access so Grep can index and search your code.