| name | code-review |
| disallowed-tools | NotebookEdit, WebSearch, WebFetch |
| description | [DEPRECATED -> use /deep-review] Systematic, bias-free code review. Renamed to /deep-review to end the literal collision with Claude Code's built-in /code-review (2.1.147+, --comment/--fix). This stub forwards to /deep-review. |
/code-review — DEPRECATED alias for /deep-review
This skill was renamed. /code-review (agent-harness) literally collided with Claude Code's
built-in /code-review command (2.1.147+, which ships its own --comment/--fix flags).
The reframed skill — which keeps the differentiators: multi-perspective specialist reviews,
adversarial cross-verification, the bias-reduction contract (anchor-free, defect-assumption,
author-neutral, context-isolated), and the structured report — lives at /deep-review,
with --comment/--fix parity behind explicit gates.
Behavior
- Print (in the user's language): "
/code-review (agent-harness) is deprecated and renamed to
/deep-review. The new skill runs plugin-shipped native Workflow segment scripts on the
workflow path (ultracode or --mode deep/thorough opt-in) and keeps the inline quick path
otherwise."
- Ask via AskUserQuestion (in the user's language):
- header: "Deprecated"
- question: "Run /deep-review with these arguments instead?"
- options:
- "Yes" / "Delegate to /deep-review with the same target/flags"
- "No" / "Cancel"
- If "Yes" → invoke the
deep-review skill with the identical arguments (target + any
--flags), delegating all behavior to skills/deep-review/SKILL.md.
- If "No" → halt.
(No orchestration logic lives here. /deep-review is the single source of truth.)
