The Agent Skills Directory

[Command Execution via Shell]: The skill uses the Bash tool to execute the jar utility, which is necessary for assembling the plugin bundle.
Evidence: Step 4.1 uses shell commands to navigate to the staging directory and execute jar cMf to create the final archive.
Security Consideration: Executing shell commands using variables (such as the plugin name) sourced from an external manifest.json file creates a potential path for command injection if the input is not strictly validated.
[Processing External Configuration]: The skill ingests and reconciles data from a JSON configuration file.
Evidence: Step 2.1 reads ppmplugin/staging/manifest.json to determine the output filename and which binaries to include in the bundle.
Security Consideration: This represents an indirect prompt injection surface. A maliciously crafted manifest could attempt to influence the agent's logic or exploit the command execution path described above, especially if the file is provided by an untrusted third party.
[User Confirmation Checkpoints]: The skill includes interactive prompts before performing potentially destructive or ambiguous actions.
Evidence: Step 2.5 uses AskUserQuestion to handle platform mismatches, and Step 4 uses it to prevent silent overwriting of existing bundles.
Context: These checkpoints are positive safety features that provide user oversight and reduce the risk of accidental data loss or incorrect builds.

assemble-ppmplugin