Skip to main content
Search helps you find the context behind previous agent work. You can search from the CLI, in Entire.io, or from your agent with the search Skill. Search combines semantic matching with keyword matching, so you can find work by meaning, exact phrases, feature names, file names, or error text. In this tutorial, you will create a small app called Launch Log, make three agent-assisted commits, and search that history.

​
Prerequisites

Before you start, make sure you have:
  • Entire installed
  • An AI coding agent installed
  • Git configured for the project
  • An Entire account
Log in to Entire:
You need to be logged in so search can return only repos and checkpoints you have access to.

​
Create a New Project

Create a Vite project:
Enable Entire for your agent. This example uses Codex:
Replace codex with the agent you use. Start your agent:

​
Build the Starter App

Ask your agent:
When the agent finishes, commit and push:
This gives you a clear phrase to search for later: mission manifest.

​
Add Readiness Scoring

Ask your agent:
Commit and push:
This gives you history that can be found by meaning, even if your search does not use the exact commit message.

​
Add an Archive

Ask your agent:
Commit and push:
This gives you another feature to search for from the CLI, Entire.io, or your agent.

​
Wait for Indexing

After your final git push, give Entire a couple minutes to index the pushed checkpoints. Then confirm you are still in an Entire-enabled repo:

​
Search by Exact Phrase

Search for the phrase you asked your agent to add:
Exact phrase searches are useful when you remember a UI label, feature name, file name, error message, or specific wording from the work.

​
Search by Meaning

Now search without using the exact commit message or heading:
Search should find the readiness scoring work because the query describes the meaning of the change, not just its exact words.

​
Search in Entire.io

You can also search from Entire.io. Go to entire.io, sign in, choose Search in the sidebar, and search for:
Search results can include related checkpoints, sessions, and commits. Open a result to inspect the captured agent work, code changes, and surrounding context.

​
Search from Your Agent

If you have Skills installed, your agent can search history for you. Open your agent and ask:
The search Skill uses entire search, summarizes likely matches, and helps your agent decide what context to inspect next.

​
Search Across Repos

If you have access to multiple Entire-enabled repos with indexed checkpoints, search across them:
Cross-repo search is useful when you remember the topic, but not which repo the work happened in.