clodcast is a single-skill Claude Code plugin released as tagged vX.Y.Z versions. Only the latest release (and the main branch) is supported — fixes ship forward, not as backports to older tags.

Please report security issues privately — do not open a public issue.

Use GitHub's private vulnerability reporting: Report a vulnerability. This opens a private advisory visible only to you and the maintainer; nothing is public until an advisory is published.

If you can't use that, email clodcast@cortech.online instead.

This is a solo-maintained project, so responses are best-effort: expect an acknowledgement within about a week. Please give a reasonable window to ship a fix before any public disclosure.

Helpful things to look at, given how the skill works:

• Credential handling. R2 credentials and the Pages deploy-hook URL must come from the environment or a 0600 ~/.config/daily-podcast/secrets.json — never from config.json or anything committed to git. A path that leaks them into logs, the run log (runs.jsonl), a workdir, or a committed file is in scope.
• Subprocess boundaries. The renderer shells out to ffmpeg/ffprobe and the save-to-spotify CLI. Argument-injection or unsanitized data reaching a shell is in scope.
• Generated output. The HTML episode description is generated from item content; unescaped, attacker-controllable input reaching that markup is in scope.
• Destructive operations. --prune-workdirs deletes directories under the temp dir; any way to make it delete outside its guarded set (symlink escape, the active workdir, N <= 0) is in scope.

• Vulnerabilities in third-party dependencies (ffmpeg, the save-to-spotify CLI, Qwen3-TTS / MLX, Python packages) — report those upstream. We'll still take a fix that changes how clodcast uses them.
• Anything requiring an already-compromised local machine or a malicious ~/.config/daily-podcast/ the operator wrote themselves.