The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted Java source files provided by the user. A sophisticated attacker could embed malicious instructions within comments, string literals, or metadata of the Java source to attempt to influence the agent's behavior during the multi-step conversion process.
Ingestion points: The skill reads .java source files to perform conversion (referenced in SKILL.md).
Boundary markers: The workflow uses a 4-step thinking process and a verification checklist (assets/checklist.md), but lacks explicit instructions to the agent to disregard instructions embedded within the source code itself.
Capability inventory: The skill instructs the agent to write new files to the disk (Step 2) and execute shell commands (Step 3) using Git.
Sanitization: There is no evidence of sanitization or filtering to detect or strip natural language instructions from the code being processed.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands to maintain Git history and manage file movements.
Evidence: SKILL.md contains explicit Git commands such as git mv src/main/java/com/example/Foo.java src/main/kotlin/com/example/Foo.kt and git commit -m "..." to be executed by the agent.

kotlin-tooling-java-to-kotlin