Prize
Best Use of Entire
Awarded to the team whose project makes the most creative use of Checkpoints, Graph or Brain.
Prize
Awarded to the team whose project makes the most creative use of Checkpoints, Graph or Brain.
Entire team
Entire is the git forge for AI-native teams. The Entire CLI captures every AI coding agent session, so the reasoning behind the code stays with the git repo, right next to your code. Then, Continuous Memory (Entire Graph + Entire Brain) turns those sessions into living knowledge for your agents. Your repo learns from every attempt, every fix, every decision, and your agents get smarter because of it.
Connect a Git commit to the agent sessions and decisions behind it.
You or your agent can search semantically through previous commits, sessions, and decisions, or literal source text, across pushed, indexed repositories.
Hands your agent a map of your codebase and can save up to 71% in token spend. Analyzes local code to find implementations, trace relationships, and investigate potential change impact.
The memory layer that lives in your repo, not in one agent. Start a session in Claude Code, pick it up in Codex, and the context carries over.
Capture work from supported coding agents, or add another through the Agent Integration Protocol.
Start with a failing test or bug, inspect the affected code with Graph, and trace relevant checkpoints to find where an incorrect assumption entered the implementation. Show the evidence behind the diagnosis.
Compare the original request with the committed change. Flag missed requirements, unexpected behavior, and claims that lack test evidence, with links to the relevant transcript and code.
Search previous sessions for similar attempts and rejected approaches, then use Graph to locate reusable code. Produce a short briefing before the agent starts editing.
Compare Graph snapshots across commits to spot new dependencies or unexpected coupling. Attach checkpoint context explaining why each change was introduced.
Let a new teammate ask, "How does authentication work, and why was it designed this way?" Combine current code relationships with historical decisions and source links.
Use a checkpoint's changes and Graph relationships to suggest behaviors that need tests. Consult session context for edge cases the agent discussed but may not have covered.
Turn a checkpoint into a concise walkthrough: the problem, key decisions, implementation, and recorded validation. Generate a shareable page with links back to the evidence.
Build an adapter that captures its sessions and connects them to commits. Ship it with lifecycle tests and a working checkpoint demo.
Feel free to be creative and impress!
Everyone on the team signs in to Entire with Sign in with GitHub and authorizes the Entire GitHub App when prompted. You can do this at any time, before or after the next step.
A GitHub organization admin installs the Entire GitHub App on the organization by following the onboarding. Nothing repository-related appears for anyone in the team until this step is done.
Sign-in order doesn't matter, but the repository and its sessions only show up after the admin has configured access.
Once the admin has imported the repository, it appears for the whole team. No invites and no permission setup: access follows the GitHub collaborator list, so everyone keeps the same roles they have on GitHub.
Next: install the CLI and start your first checkpoint (following section).
# macOS; see Resources for other platforms
brew install --cask entireio/tap/entire
cd your-repo
entire enable --agent codex
Replace codex with your agent's integration name. Start a new agent session, ask it to make a focused change, and review the result.
git add path/to/changed-file
git commit -m "Build my first feature with Entire"
Choose yes if prompted to link the commit to session context.
entire checkpoint list
entire checkpoint explain HEAD
A normal committed checkpoint needs both captured agent context and a Git commit. Full walkthrough
You'll see Trails in action during our demo. To use it in your own repository:
Trails is behind a feature flag. Once you've signed up, fill in the Trails access form and someone from the Entire team will enable Trails for you.
Open your repository in Entire, go to Settings → General, and turn on Runners, Auto build on push, and Trails. This requires admin permission on the repository.
Runners evaluate each Trail and publish findings. From the repository root, run:
entire runner setup -y
Then commit and push the generated .entire/ files.
Your coding agent already knows how to use Trails through the CLI, so just ask it to create one. Or run:
entire trail create
Commit and push to the Trail branch as usual. The Trail updates automatically, Runners evaluate each push, and your team can review and merge from the Trail.
Optional: only if you are building with Entire Graph/Brain.
You or your agents can install the entire-graph plugin. Activate it once per repository:
entire plugin install graph
entire graph init-agents --repo .
The second command creates .entire/graph-agent.md with the agent instructions, and adds a short block to your AGENTS.md and CLAUDE.md telling your agents to use entire-graph.
Commit those three files so it applies to your whole team, then start a fresh agent session. You'll know it worked when your agent's first move is entire-graph search instead of a grep.
Restart the agent so it picks up the new instructions, then ask in plain language: Where is request routing implemented?
The agent runs graph queries, reads the code the graph points to, and answers with citations.
To see what it saved you, run:
entire graph stats --repo .
It reads your Claude Code session transcripts, counts how many times your agent used the graph instead of grepping and reading files, and estimates the tokens it saved. For how entire-graph works, installation, and troubleshooting, see the Graph docs .
Make sure the Entire CLI and entire-graph are installed first (steps above).
curl -fsSL https://raw.githubusercontent.com/entireio/entire-brain/main/scripts/get-brain.sh | bash
The script downloads a prebuilt binary, checks it against the release checksums, and registers it with the CLI as the entire-brain plugin.
Needs Go 1.27+, Git 2.36+, the Entire CLI on your PATH, and a C compiler for Graph's tree-sitter bindings.
git clone https://github.com/entireio/entire-brain.git
cd entire-brain
./scripts/install.sh
This builds and registers entire-brain and entire-graph. It does not initialize a repository's memory or install the watcher. Then cd to the project repository where you want Brain and run:
entire brain setup
From your project repository, install the agent instructions:
entire brain init-agents
This embeds your existing sessions (if any) and creates or updates the agent instructions at .entire/agent-guide.md. Rerunning the command replaces this file. It also updates AGENTS.md and CLAUDE.md with references to the guide inside managed blocks; your text outside those blocks is preserved.
Review the generated files and commit them together when the instructions should apply to your team. When entire-graph is installed, the guide tells agents to use entire-brain for prior decisions and entire-graph to locate and inspect code. Restart your agent or reload its repository instructions so it picks up the guide.
Print this repository's configuration:
entire brain mcp --print-config
Then register it with an MCP-capable client. For example, in Claude Code:
claude mcp add entire-brain -- entire brain mcp
See the agent integration guide and coordination contract for more setup details.
Suggested submission checklist
entire status to verify checkpoint synchronization.For GitHub-backed repositories, share:
https://entire.io/gh/OWNER/REPO/commit/FULL_COMMIT_SHA