The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes jj (Jujutsu) CLI commands to retrieve source code diffs and file contents. This behavior is essential for the stated purpose of performing code reviews and does not involve elevated privileges or suspicious arguments.
[INDIRECT_PROMPT_INJECTION]: The skill reviews untrusted external code, which introduces a surface for indirect prompt injection where malicious instructions in the code could target the review sub-agents.
Ingestion points: Local file content and version control diffs retrieved via jj commands in SKILL.md and SUBAGENT.md.
Boundary markers: Absent; file content and diffs are interpolated directly into prompts for specialized sub-agents (e.g., quality-reviewer, security-researcher).
Capability inventory: Executes jj CLI commands and spawns specialized sub-agents via the Task tool.
Sanitization: Absent; the skill passes code content to sub-agents without filtering or escaping.

code-review