Skip to content

Security: segmently-ai/segmently-modules

SECURITY.md

Security policy

Supported versions

Security fixes go into the latest 0.x minor release.

Reporting a vulnerability

Report it privately, never in a public issue:

  • GitHub's Report a vulnerability button on the Security tab of this repository (private vulnerability reporting), or
  • email leonid@segmently.ai.

You get an acknowledgement within 5 working days. Please include the version (segmently-modules --version), what you ran and what happened.

Trust model

  • The command line runs locally, with the rights of the user who runs it. It makes no network call itself; the one exception is init finish --open-pr (and init resume --open-pr), which runs git push origin and gh pr create for the branch it made.
  • A repository can run commands of its own: a [[hooks.post_sync]] hook in its team config or in a pack it installs runs on segmently-modules sync with your environment (sync --dry-run runs none), the way a build script of that repository runs. Review a repository's .segmently-modules/ before you run sync in it, as you would review its build scripts.
  • The agent command (SEGMENTLY_MODULES_AGENT_COMMAND) is read only from your environment, never from a config file. A [match] mode that a config file sets runs it only when you also set SEGMENTLY_MODULES_AGENT_ALLOW_REPO_MODE=1; --match and choose --run are your own requests. Details: docs/configuration.md.

There aren't any published security advisories