The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from the project's commit history to analyze patterns and propose scopes. This creates an indirect prompt injection surface where a specially crafted commit message could attempt to influence the agent's logic during Phase 2 or 3.
Ingestion points: git log --format="%s" -n 50 output in Phase 2.
Boundary markers: The prompt instructions do not provide delimiters or instructions to ignore embedded commands within the log data.
Capability inventory: Bash, Write, Read, Glob.
Sanitization: No sanitization or validation of the commit message content is performed before analysis.
[COMMAND_EXECUTION]: The skill uses shell commands to inspect the local environment and modify Git configuration. It executes git config to set user identity and ls/find to determine project structure. These operations are limited to the project directory and standard Git settings.

config-git