Only the latest release of flow-next receives security fixes. The plugin runs locally inside your agent harness (Claude Code / Codex / Droid) with no hosted components — the attack surface is the bundled flowctl CLI, the shell scripts, and the skill/hook prompts.
| Version | Supported |
|---|---|
| latest release | ✅ |
| older releases | ❌ — upgrade first |
Please do not open a public issue for security problems.
- Preferred: GitHub private vulnerability reporting
- Or email: gordon@mickel.tech
Include reproduction steps and the affected version (flowctl --version or the README badge). You'll get an acknowledgment within a few days; fixes ship as a normal release with credit unless you prefer otherwise.
- Prompt-injection hardening of skills (e.g. Ralph guardrails, review gates) is in scope — flow-next's autonomous loops are designed to be gated and receipted, and bypasses are bugs.
- Vulnerabilities in host platforms (Claude Code, Codex, Droid, Cursor, Grok) or optional third-party tools (
clawpatch,lavish-axi, RepoPrompt) belong upstream — but if flow-next's integration makes them worse, report here too.