The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads from ~/.ai-review/credentials.json. While this is intended to verify the setup of the ai-review CLI, it exposes the existence and contents of a sensitive credentials file to the agent's context.
[EXTERNAL_DOWNLOADS]: The skill instructs users to install an external package @zawlinnnaing/ai-review-cli via npm. This is a vendor-provided resource, but it represents an external dependency required for the skill's operation.
[COMMAND_EXECUTION]: The skill uses the bash tool to execute multiple CLI commands including ai-review create-mr, ai-review get-context, and ai-review post-description. These commands use arguments derived from user input and external file content.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources (GitLab MR diffs) to generate automated summaries.
Ingestion points: Data is read from ~/.ai-review/mr-context.json after fetching MR details.
Boundary markers: No specific delimiters or instructions are provided to the agent to ignore potentially malicious instructions embedded within the code diffs.
Capability inventory: The agent has access to bash for command execution and file writing.
Sanitization: There is no evidence of sanitization or filtering applied to the MR context before the agent processes it to write the description.

create-mr