The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides bash scripts and GitHub CLI (gh) commands for the agent to execute locally. These scripts use git grep, git tag, and git fetch to analyze the repository's history and release state.
[INDIRECT_PROMPT_INJECTION]: The bash scripts provided in Step 1 use shell variables (PROC and FLOOR) that are interpolated directly into command strings. This creates a potential surface for command injection if these variables are populated with malicious payloads.
Ingestion points: Shell variables PROC and FLOOR defined within the bash blocks in SKILL.md.
Boundary markers: No delimiters or safety instructions are present to prevent the agent from executing injected commands within these variables.
Capability inventory: The skill assumes access to a local shell environment with tools like git, sed, awk, and sort.
Sanitization: There is no evidence of input validation, escaping, or sanitization applied to these variables before they are executed in the shell context.

trpc-compat