The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to perform its core functions.
It runs which stagereview to check for the presence of the required CLI tool.
It runs git rev-parse --is-inside-work-tree to verify the execution context is a git repository.
It executes stagereview prep to generate diff data and stagereview show to launch a local review interface.
[EXTERNAL_DOWNLOADS]: The skill instructions advise the user to install the stagereview package globally via npm (npm install -g stagereview) if the command is not found. This is a legitimate dependency for the skill's functionality.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it processes untrusted data from the local environment.
Ingestion points: The skill reads git commit messages and diff hunks from a temporary file generated by the prep command.
Boundary markers: While the data is delimited by headers like === COMMIT MESSAGES === and === HUNKS ===, there are no specific instructions for the agent to ignore or escape instructions that might be embedded within these git objects.
Capability inventory: The skill possesses the ability to execute shell commands and write to the file system via mktemp and cat heredocs.
Sanitization: The instructions do not specify any sanitization or validation of the content retrieved from the git repository before it is processed by the LLM.

stage-chapters