The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs remote command execution via ssh to a hardcoded internal IP address (192.168.0.15) to install packages, rebuild native modules, and manage processes using PM2.
[PRIVILEGE_ESCALATION]: The deployment instructions explicitly mandate the use of the root user when executing remote commands and transferring files via scp (root@192.168.0.15).
[DYNAMIC_EXECUTION]: The skill uses node -p to dynamically evaluate JavaScript snippets for version extraction from package.json and executes arbitrary commands on a remote host via ssh strings.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from local files without proper sanitization or boundary markers.
Ingestion points: Reads content from CHANGELOG.md, package.json, and docs/reference/openapi.yaml (Step 9 and Step 13).
Boundary markers: None identified; the skill directly pipes extracted text into shell command arguments.
Capability inventory: The skill possesses capabilities for remote command execution (ssh), repository modification (git push), Pull Request creation (gh pr create), and package registry publishing (npm publish).
Sanitization: No sanitization is performed on the ingested text beyond basic whitespace trimming using sed.

generate-release-cx