Tattlr
Reads the logs AI coding agents leave on disk, writes a rap sheet, plants trap credentials, and can block bad calls through tattlr protect.
Open source
Tattlr
A rap sheet for AI coding agents
Tattlr reads the logs AI coding agents already leave on disk and tells you what they actually did. It can plant trap credentials, and tattlr protect can stop a bad call before it runs. Nothing leaves the machine.
What it does
Logs, traps, and a live block
- Works with Claude Code, Codex CLI, Gemini CLI, OpenClaw and Cursor.
- tattlr prints a rap sheet: sessions, tool calls, and the worst moments.
- tattlr trap plants fake keys. An agent has no reason to read them.
- tattlr protect hooks into the agent and can block the call before it runs.
- Local only. No account, no telemetry, no network calls. MIT license. Written in Go, for macOS and Linux.
Install
Pick one
Each of these installs the same program.
- Homebrew
brew install kaygaaf/tap/tattlr - npm
npm i -g tattlr - Go 1.22 or newer
go install github.com/kaygaaf/tattlr/cmd/tattlr@latest - Install script
curl -fsSL https://raw.githubusercontent.com/kaygaaf/tattlr/main/install.sh | sh
The install script is short. Read it before you pipe it into a shell. Releases and checksums are on GitHub.
Open source
Public repositories
Tattlr is first. The others are separate tools on the same GitHub account.
busy-port
Shows which process is using a TCP or UDP port.
compose-env
Checks docker-compose files for missing ${ENV} variables without printing secrets.
ha-yaml-kit
Helpers for Home Assistant automation YAML.
dotenv-guard
A CLI and pre-commit hook that blocks .env commits and scans for secret patterns.