The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Python script (scripts/codeex_plugin.py) to execute shell commands via the subprocess.run module. These commands are used to invoke the Node.js binary for syntax checking (node --check) and to execute management scripts within the target repository (scripts/plugins-cli.mjs). While these are functional requirements for the skill's purpose as a development tool, they represent an execution surface for local code.
Evidence: scripts/codeex_plugin.py contains subprocess.run([node, "--check", str(entry)], ...) and subprocess.run([node, "scripts/plugins-cli.mjs", *args], ...).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to read, analyze, and modify existing plugin source code and plugin.json manifest files within a user-provided project root. This creates an attack surface where a malicious project or plugin could contain instructions designed to influence the agent's behavior during the analysis or implementation phase.
Ingestion points: The agent is instructed to "Inspect the current hook call sites," "Search existing plugins for the closest contract example," and "Inspect the adjacent native DOM" (SKILL.md Steps 2, 4, and 5).
Boundary markers: The instructions lack explicit boundary markers or warnings to disregard instructions embedded within the analyzed plugin source code.
Capability inventory: The skill has the capability to write files, rename directories, and execute shell commands (node) via the scripts/codeex_plugin.py utility.
Sanitization: The skill performs basic validation on plugin IDs, versions, and permissions using regular expressions, and ensures entry paths remain within the plugin directory.

lov-codeex-creator