Causari
Count how many AI-tagged lines survive in a git repo and log agent prompts per edit
- Pricing
- Open source
- Platforms
- CLI, macOS, Linux
- Runs
- Locally
- Version
- 0.5.0
- Listing updated
- 2 days ago
About Causari
Causari is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.
Causari answers a narrow question: of the lines introduced by commits whose git metadata matches an AI rule, how many does git blame still attribute to those commits at HEAD? The re audit command reads Co-Authored-By trailers, bot authors and agent markers, and reports a survival count next to a baseline of untagged lines in matched age windows. It works on any git repository, including a public one passed by name.
The tool is blunt about what it cannot see. Inline completions leave no git trace, commits without a trailer are unknown rather than human, and a metadata match does not prove a model wrote the line. A one-off bulk commit can dominate the ratio, so it also reports a capped ratio and a per-commit median.
The second half is a local ledger. Hooks for Claude Code and Cursor record each prompt, model and file touched into a gitignored .causari/ folder, and re why file:line names the event linked to a line. This ledger is what the MCP server (re mcp) exposes. It is one Rust binary, Apache 2.0, with no account or telemetry.
Install
Run in a terminal
claude mcp add causari -- npx -y causariAdd to claude_desktop_config.json
{
"mcpServers": {
"causari": {
"command": "npx",
"args": [
"-y",
"causari"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"causari": {
"command": "npx",
"args": [
"-y",
"causari"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"causari": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"causari"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"causari": {
"command": "npx",
"args": [
"-y",
"causari"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"causari": {
"command": "npx",
"args": [
"-y",
"causari"
]
}
}
}What it does
- Survival audit
- Counts AI-tagged lines still alive at HEAD using git blame.
- Age-matched baseline
- Compares tagged lines to untagged lines in the same repository by age.
- Agent edit ledger
- Records prompt, model and files for each agent edit in a local folder.
- Line lookup
- re why names the ledger event associated with a line.
- Signed seals
- Authenticate that a key signed a measurement for a commit and method.
- Single binary
- One Rust binary with no cloud, account or telemetry.
Good for
- Measure how much AI-assisted code from a repo is still alive
- Add a survival count to CI as Markdown
- Trace which prompt produced an edit
- Compare an agent's code survival against human commits
Pricing
Open source
Apache 2.0, no account
Prices change. Check causari.dev before you buy.
Causari FAQ
Does it prove who wrote a line?
No. AI-tagged means metadata matched; the ledger records what the runtime declared, not authorship.
What does it miss?
Inline completions such as Copilot or Cursor Tab leave no git trace, and untagged commits count as unknown.
What is the licence?
Apache 2.0.
Does it need setup?
The audit runs with one command on any git repo; the ledger needs re init and a hook.
Is Causari free?
Yes. Causari is open source, so you can use and self-host it for free.
How do I add Causari to Claude Code?
Run this command in your terminal: claude mcp add causari -- npx -y causari
What are the best alternatives to Causari?
The closest alternatives to Causari are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.
Alternatives to Causari

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools

Codebase Memory
MCP serverLocal MCP server that builds a code knowledge graph for AI coding agents
Runs locally- Open source
- 48.7K GitHub stars
- Knowledge and memory

GitHub
MCP serverOfficial MCP server connecting AI assistants to GitHub repos, issues and workflows
Not checked yet- Remote
- Open source
- 34.5K GitHub stars
- Developer tools

Serena MCP: the IDE for your agent
MCP serverMCP toolkit giving coding agents symbol-level code retrieval and editing
Runs locally- Open source
- 31.5K GitHub stars
- Developer tools

XERJ
MCP serverLocal search engine for AI agents with cited answers over your code and docs
Runs locally- Open source
- 19.2K GitHub stars
- Knowledge and memory

Agent-Native Analytics
MCP serverOpen-source analytics app where an AI agent builds charts and dashboards from your data
- Remote
- Open source
- 7.7K GitHub stars
- Analytics and data