Only the latest minor (currently 0.8.x) receives security fixes. Older versions are not maintained.

Use GitHub's private security advisory flow to report anything that affects:

• API key storage / leakage (the ~/.config/ia-pptx/credentials.json path, process-list visibility, environment leakage)
• Carte blanche execution scope (a path where Claude Code could run arbitrary commands the user did not explicitly authorize)
• Subprocess argument handling (gen_image.py, freeform_helpers, pptxgenjs / WeasyPrint invocations)
• Any path that downloads remote resources at install/run time

Do not open a public issue for security-relevant reports. We aim to acknowledge within 72 hours and ship a fix within 7 days for high-severity issues.

• Carte blanche is opt-in (default ON in Streamlit, can be disabled). When enabled, Claude Code has Bash/Read/Write/Edit access in a tmp cwd. Users who don't want this can disable it; the tools are then narrowed to Read (or Read+Bash if Nano Banana is on).
• API keys live at ~/.config/ia-pptx/credentials.json (mode 0600 on POSIX). They are loaded into the Python process and passed to subprocess calls. They are never logged, but they are visible via ps aux while a child process is running with them in argv (this is the API server's design — we don't pass keys via argv ourselves).
• The skill bundle runs untrusted Claude code via Node + LibreOffice
  • WeasyPrint subprocesses. None of these have network ACLs by default.