The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests git commit history via git log and git show and delegates analysis to a subagent via the Agent tool. A malicious commit message could potentially influence the subagent's analysis or the final proposal.\n- Ingestion points: Git commit messages and repository history gathered in Step 1 of SKILL.md.\n- Boundary markers: The instructions do not specify delimiters or prompts to ignore instructions embedded within commit messages.\n- Capability inventory: The skill has the ability to write to the file system (Write, Edit) and execute shell commands (Bash), though these are gated by a user approval step (AskUserQuestion).\n- Sanitization: There is no evidence of sanitization or filtering of the commit data before it is processed by the agent or subagent.

sync-claude-md