Photoshop Automator

Automate Adobe Photoshop on Windows via ExtendScript to run scripts, update text layers, create layers, apply filters, play actions, and export images.

This skill legitimately runs user-provided ExtendScript in Photoshop and therefore can read and write local files — that is the intended functionality but also a real risk. Only install/use it if you trust the skill and the scripts you run. Before enabling or allowing autonomous use: 1) Verify the skill source (registry lists 'unknown' homepage; package files reference a GitHub repo—confirm the upstream repo and author). 2) Do not run unreviewed JSX; inspect any generated or supplied scripts. 3) Consider restricting agent autonomous invocation for this skill (disable autonomous runs) so scripts only run with your explicit approval. 4) Be aware export paths and temporary files may overwrite or leave artifacts; run initially in a safe/test environment (or VM) and limit file paths and permissions where possible. 5) Notify maintainers to fix metadata that requires both cscript and osascript unconditionally, since that is incorrect for typical single-OS hosts.

Type: OpenClaw Skill Name: photoshop-automator Version: 1.2.4 The skill is classified as suspicious due to a significant Remote Code Execution (RCE) vulnerability. The `runScript` command in `handler.js` directly executes arbitrary ExtendScript (JSX) code provided by the user/agent, which, as explicitly stated in `SKILL.md`, has 'direct access to the host filesystem' and 'can create, modify, or delete files'. While the `SKILL.md` warns the AI to 'Always review dynamically generated scripts', this capability makes the skill highly susceptible to prompt injection attacks, allowing a malicious user to execute arbitrary code on the host machine with filesystem access. This is a severe vulnerability, not intentional malice by the skill itself, as the skill's purpose is to provide powerful Photoshop automation.