The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the current conversation and command-line arguments to construct a Git commit message, which is then executed in a shell.
Ingestion points: The workflow resolves an issue key and constructs a commit subject using $ARGUMENTS and the current conversation context (SKILL.md).
Boundary markers: The skill lacks instructions to delimit or escape the content used in the commit subject to prevent it from being interpreted as shell commands.
Capability inventory: The skill utilizes shell-based tools including git add and git commit (SKILL.md).
Sanitization: No sanitization or validation logic is present to filter potentially malicious characters (e.g., backticks, semicolons, or command substitutions) that might be injected via the conversation or arguments.
[COMMAND_EXECUTION]: The skill executes standard Git commands. While these are routine operations, the direct interpolation of user-controlled text into the git commit -m command presents a risk if the execution environment does not provide automatic argument escaping.

fast-commit