Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed local Excalidraw whiteboard helper, but installing it runs a pinned community runtime build and it keeps local board servers and saved board data across sessions.
Install only if you are comfortable with a local tool that downloads and builds a pinned community Excalidraw runtime, stores boards under ~/.local/share/codex-whiteboard, and runs unauthenticated loopback-only servers for active boards. Keep it off exposed network interfaces and avoid putting secrets on shared whiteboards unless you accept local persistence.
Referenced artifact was not completely inspected
Resolve `scripts/session.mjs` relative to this `SKILL.md`. Use its absolute path so commands work from any project. Examples below use `SESSION_CLI` for that pa
Referenced artifact was not completely inspected
Resolve `scripts/session.mjs` relative to this `SKILL.md`. Use its absolute path so commands work from any project. Examples below use `SESSION_CLI` for that pa
Referenced artifact was not completely inspected
y under `~/.local/share/codex-whiteboard/`. If it is missing, run this skill's `scripts/setup.mjs`; it builds the pinned source and requires network access, Git
The skill invokes local shell commands, uses environment variables, and explicitly requires network access during setup, but it declares no tool scope or permission boundary. That creates an authorization gap where an agent may perform sensitive actions without an explicit least-privilege contract, increasing the chance of unintended command execution, environment access, or downloads.
The skill intentionally persists whiteboard state across chats and process restarts and launches loopback servers for ongoing sessions. Persistent local state can expose prior project data to later sessions or other local processes/users, and long-lived local services expand the attack surface if session isolation, file permissions, or browser access are weak.
metadata:
author: Jim Jeffers
version: "0.1.0"
compatibility: "Requires a local shell, Node.js 20+, npm, Git, and a browser on the same computer. Setup downloads and builds a pinned runtime; sessions write local state and launch loopback servers."
openclaw:
homepage: https://github.com/jimjeffers/excalidraw-whiteboard-skill
requires:
The added endpoint exposes internal identifiers and the process ID to any caller without authentication or clear necessity. While these values are not secrets by themselves, they provide environment and process metadata that can aid reconnaissance, correlation of sessions, and targeting of the running service, especially if the whiteboard server is reachable beyond a strictly local trust boundary.
The setup script performs a live bootstrap from a remote GitHub repository and then builds and installs that code locally. Even though it pins a revision and verifies the fetched HEAD matches that revision, this still introduces network-sourced code execution during setup (git, npm ci, npm run build), which exceeds the declared local whiteboard functionality and creates a supply-chain and install-time execution risk.
The repository bootstrap and package installation are network-dependent and trigger package-manager lifecycle behavior on downloaded content. For a skill described as editing local Excalidraw whiteboards, this is broader than necessary and increases exposure to dependency confusion, compromised upstream packages, malicious install scripts, or unexpected code execution during environment preparation.
Detected: suspicious.dangerous_exec