CodeGuard already supports code review. A subagent would not add a new capability. It would change how a review is executed.

The main benefits are operational:

• Context isolation: Review work stays out of the main conversation.
• Parallel execution: Multiple files or pull requests can be reviewed at the same time.
• Independent verification: Review is performed by a separate agent instead of the one that generated the code.
• Specialized configuration: The reviewer can use its own prompt, tools, and constraints without affecting the main agent.

See Cursor’s guidance on when to use subagents.

Multi-IDE support

Subagents are now supported in several AI coding tools, which makes this a reasonable fit for the existing translator pipeline.

Examples:

• Claude Code: .claude/agents/*.md with YAML frontmatter (docs)
• Cursor: .cursor/agents/ directory (docs)
• OpenCode: .opencode/agents/*.md with mode: subagent (docs)

These formats are similar enough that a new translator in src/formats/ could generate agent definitions alongside the rule files. The metadata differs by tool, but the overall structure is close enough to support.

SIG alignment

The SIG scope document (Deliverable 3) calls for an "AI-on-AI review architecture using CodeGuard rules." A dedicated reviewer subagent with independent context would be a direct implementation of that idea.

Open questions

1. Is the context-isolation benefit worth the maintenance cost of a second artifact? The skill already handles review, so the subagent would be an optimization rather than a new capability.
2. What should the initial scope be? Should the first version focus only on rule-compliance review, or should it also orchestrate external scanners such as Semgrep, CodeQL, and osv-scanner?
3. What should the output format be? Options include SARIF for GitHub Security tab integration, a Markdown report, inline pull request comments, or configurable output.
4. Should we have one subagent or a small catalog? Possible follow-ons include @threat-modeler for STRIDE-style analysis on architecture changes and @dependency-auditor for SBOM and CVE review.

Prior art

• Anthropic’s built-in /security-review command (Claude-only)
• GitHub Copilot Autofix (GitHub-only)