Agentry is pre-1.0 and ships as a Claude Code plugin. Security fixes target the latest release and main.

Please do not open a public issue for security problems.

Report privately via GitHub's Security Advisories (Security → Report a vulnerability). Include:

• a description and impact,
• steps to reproduce or a proof of concept,
• affected version / install method (marketplace, release zip, local),
• any suggested fix.

You can expect an acknowledgement within a few days and a coordinated fix before public disclosure.

Agentry runs inside Claude Code and ships markdown (agents/commands/skills) plus dependency-free Node scripts behind one command, agentry (plugin/bin/agentry). Areas worth scrutiny:

• the memory CLI (agentry mem, plugin/scripts/lib/memory) — it reads/writes files under the resolved memory roots and runs a local node:sqlite index; path handling and untrusted-input parsing matter,
• the lane scripts (plugin/scripts) — dependency-free Node that the commands run to create worktrees, run the repo's gate and land commits,
• anything that handles untrusted repo content a conductor or agent might read.

Out of scope: vulnerabilities in Claude Code itself (report those to Anthropic), or in third-party tools/MCPs a user has configured separately.