delegating-code-review

name	delegating-code-review
description	Use when you have made code changes and want another CLI agent to review them before committing or continuing. Covers crafting review prompts, scoping reviewer permissions to read-only, interpreting feedback, and follow-up clarification sessions.

Delegating Code Review to Another Agent

Use AgentShell to invoke another CLI agent to review changes in a repository. The reviewing agent has its own tools — just point it at the repo, tell it what to look at, and let it do the work.

Assumes familiarity with AgentShell basics. See invoking-cli-agents for setup and core API.

When to Use

• You have made code changes and want a second opinion before committing
• You want to validate that changes meet requirements before marking work complete
• You need a security, performance, or correctness review
• You want to check for regressions or unintended side effects

Review Uncommitted Changes

Tell the reviewer to look at the current uncommitted changes in the working directory.

from agent_shell.shell import AgentShell
from agent_shell.models.agent import AgentType

reviewer = AgentShell(agent_type=AgentType.CLAUDE_CODE)

review = await reviewer.execute(
    cwd="/path/to/project",
    prompt="""Review the uncommitted changes in this repository.
Focus on correctness, security, and design.
Flag issues by severity: CRITICAL, WARNING, or SUGGESTION.""",
    allowed_tools=["Read", "Glob", "Grep", "Bash"],
    model="sonnet",
)

The reviewer will run git diff itself, read surrounding code for context, and report findings.

Review a Specific Commit or Range

Point the reviewer at a particular changeset.

review = await reviewer.execute(
    cwd="/path/to/project",
    prompt="""Review the changes in commit abc1234.
Focus on correctness, security, and design.
Flag issues by severity: CRITICAL, WARNING, or SUGGESTION.""",
    allowed_tools=["Read", "Glob", "Grep", "Bash"],
    model="sonnet",
)

# Review a range of commits
review = await reviewer.execute(
    cwd="/path/to/project",
    prompt="""Review all changes between main and HEAD.
Focus on correctness, security, and design.
Flag issues by severity: CRITICAL, WARNING, or SUGGESTION.""",
    allowed_tools=["Read", "Glob", "Grep", "Bash"],
    model="sonnet",
)

Follow Up

Use session resumption to ask the reviewer to clarify or elaborate.

clarification = await reviewer.execute(
    cwd="/path/to/project",
    prompt="Can you explain the security concern in more detail and suggest a specific fix?",
    allowed_tools=["Read", "Glob", "Grep", "Bash"],
    model="sonnet",
    session_id=review.session_id,
)

Cross-Agent Review

Use a different agent or model than the one that wrote the code for genuine independence.

Safety note: Only Claude Code respects allowed_tools. OpenCode ignores it — the agent has access to all tools regardless. When using OpenCode as a reviewer, instruct it not to modify files in the prompt.

# Review with OpenCode using a different model
reviewer = AgentShell(agent_type=AgentType.OPENCODE)

review = await reviewer.execute(
    cwd="/path/to/project",
    prompt="""Review the uncommitted changes in this repository. DO NOT modify any files.
Focus on correctness, security, and design.
Flag issues by severity: CRITICAL, WARNING, or SUGGESTION.""",
    model="github-copilot/gpt-5.4",
)

Prompt Patterns

General Review

Review the uncommitted changes in this repository.
Focus on correctness, security, and design.
Flag issues by severity: CRITICAL, WARNING, or SUGGESTION.

Focused Security Review

Review the uncommitted changes for security vulnerabilities only. Check for:
- SQL injection, XSS, command injection
- Authentication/authorisation gaps
- Secrets or credentials in code
- Unsafe deserialization

Requirements Validation

The requirement was: "{original_requirement}"

Review the uncommitted changes and assess whether they fully satisfy the requirement.
Identify any gaps, missing edge cases, or partial implementations.

Test Coverage Review

Review the uncommitted changes and identify test scenarios that are missing.
For each gap, describe the test case that should exist and why it matters.

Common Mistakes

Mistake	Fix
Not checking for empty responses from execute()	execute() drops error events — an empty review.response likely means the agent failed. Use stream() if you need failure detection.
Manually capturing diffs and passing them in the prompt	Let the reviewer run git diff itself — it has tools
Using OpenCode and assuming allowed_tools works	OpenCode ignores tool restrictions — use prompt instructions or use Claude Code
Reviewing with the same model that wrote the code	Use a different model or agent type for independence
Ignoring the review and committing anyway	At minimum, address all CRITICAL items before proceeding
Not giving the reviewer Bash access	Without Bash, the reviewer can't run git diff or git log to inspect changes