The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local shell commands including git for change detection and repository management, and jq for configuration parsing. These are standard operations for a development-focused agent.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted content from project source code and local configuration files. However, this is inherent to its function as a code analysis tool and is managed by standard agent safety protocols.
[SAFE]: The skill operates entirely within the local development environment as described, with no evidence of remote code execution or unauthorized data exfiltration.

candid-improve-implementation