Skip to content

Releases

v0.3.0

Choose a tag to compare

@github-actions github-actions released this 14 Nov 13:34

Three new rules in this release.

infra-destroy asks before an agent tears down infrastructure: terraform or pulumi destroy, kubectl delete namespace, wiping an S3 bucket, and the usual cloud "delete this" commands.

imds blocks requests to cloud instance metadata services such as 169.254.169.254, metadata.google.internal, ECS/EKS metadata and Alibaba's 100.100.100.200. AWS time sync at 169.254.169.123 is left alone.

container-escape asks before a container run that gets the host: --privileged, a mounted Docker socket, -v /:/host, --pid=host, --cap-add SYS_ADMIN, or nsenter -t 1. Ordinary project mounts like -v $PWD:/src are fine.

Install

curl -fsSL https://raw.githubusercontent.com/kaygaaf/tattlr/main/install.sh | sh
brew install kaygaaf/tap/tattlr
go install github.com/kaygaaf/tattlr/cmd/tattlr@v0.3.0

Binaries for macOS, Linux and Windows are below, with sha256 sums in checksums.txt.

The full list of changes is in CHANGELOG.md. If something looks off, open an issue. If you find a security problem, please report it privately through the Security tab.

v0.2.0

Choose a tag to compare

@github-actions github-actions released this 11 Nov 06:31

tattlr scan can now write SARIF (--format sarif), so findings can go straight into GitHub code scanning. There's also --format markdown for CI job summaries and issues.

To run it in GitHub Actions after your agent step, there's a new action: kaygaaf/tattlr-action. It installs this release, checks its sha256, writes the rap sheet to the job summary and can fail the job with fail-on: high.

- uses: kaygaaf/tattlr-action@v1
  if: always()
  with:
    fail-on: high

Install

curl -fsSL https://raw.githubusercontent.com/kaygaaf/tattlr/main/install.sh | sh
brew install kaygaaf/tap/tattlr
npm i -g tattlr
go install github.com/kaygaaf/tattlr/cmd/tattlr@v0.2.0

Binaries for macOS, Linux and Windows are below, with sha256 sums in checksums.txt.

The full list of changes is in CHANGELOG.md. If something looks off, open an issue. If you find a security problem, please report it privately through the Security tab.

v0.1.1

Choose a tag to compare

@github-actions github-actions released this 06 Nov 13:19

go install and tattlr version now print the version without a leading v. tattlr scan --card file.png writes a PNG without needing extra tools, and the card shows the overall score.

On Windows arm64, npm i -g tattlr downloads the binary from this release (npm's filter blocks that platform package name).

Install

curl -fsSL https://raw.githubusercontent.com/kaygaaf/tattlr/main/install.sh | sh
brew install kaygaaf/tap/tattlr
npm i -g tattlr
go install github.com/kaygaaf/tattlr/cmd/tattlr@v0.1.1

Binaries for macOS, Linux and Windows are below, with sha256 sums in checksums.txt.

The full list of changes is in CHANGELOG.md. If something looks off, open an issue. If you find a security problem, please report it privately through the Security tab.

v0.1.0

Choose a tag to compare

@github-actions github-actions released this 29 Oct 14:53

First release.

tattlr reads the logs your AI coding agents already leave on disk (Claude Code, Codex CLI, Gemini CLI, OpenClaw and Cursor) and tells you what they actually did: secrets they read, destructive commands, stuff piped from curl into a shell, data sent out. It can also hook into those agents and block the bad calls before they run. Nothing leaves your machine.

tattlr scan                # what did my agents do?
tattlr trap                # plant fake keys as bait for prompt injections
tattlr protect install     # block the worst of it live, shows a diff and asks first

Install

curl -fsSL https://raw.githubusercontent.com/kaygaaf/tattlr/main/install.sh | sh

or with Go 1.22+:

go install github.com/kaygaaf/tattlr/cmd/tattlr@v0.1.0

Homebrew and npm are coming soon. Binaries for macOS, Linux and Windows are below, with sha256 sums in checksums.txt.

The full list of changes is in CHANGELOG.md. If something looks off, open an issue. If you find a security problem, please report it privately through the Security tab.