Agent skill

Strands Review

by strands-agents in strands-agents/harness-sdk

Local preview of the strands-agents/devtools /strands review agent.

Apache-2.0Auto-check passedDevelopment

Install Strands Review

skills CLI
$ npx skills add strands-agents/harness-sdk --skill strands-review -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install strands-agents/harness-sdk strands-review --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/strands-agents/harness-sdk.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/strands-review .claude/skills/strands-review && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
strands-review
GitHub stars
8.8k
Token cost
~3.2k tokens
SKILL.md length
1,521 words
Files
1
Skills in repo
14
Repo updated
First seen
Licence
Apache-2.0

At a glance

Local preview of the strands-agents/devtools /strands review agent.

  • Works in 6 steps: Setup Review Environment → Analyze Pull Request Context → Code Analysis Phase → …
  • The user types /strands-review
  • SKILL.md covers Role, Steps, Review Focus Areas and Best Practices, plus 1 more section
  • Calls gh

What it does

Strands Review is an agent skill from strands-agents/harness-sdk. Local preview of the strands-agents/devtools /strands review agent. Body is the upstream Task Reviewer SOP verbatim — do not paraphrase. Use when the user types /strands-review, asks for a "strands review" of a PR, or wants to anticipate what the remote /strands review GitHub Action will flag. Findings are close but not identical to the remote agent. Strongly prefer running this skill in a fresh-context subagent rather than inline — the SOP is long and reviewer judgment is more reliable when it isn't entangled…

Its SKILL.md is about 3.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Development, covering Operations and SOPs, CI/CD and Pull requests. The repository describes itself as: Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud. The licence is Apache-2.0.

When your agent uses it

  • The user types /strands-review
  • Asks for a strands review of a PR
  • Wants to anticipate what the remote /strands review GitHub Action will flag

Example prompts

  • “strands review”
  • “t entangled with the parent conversation”
  • “/strands-review”

Workflow steps

6 steps, taken from the step headings in SKILL.md.

  1. Setup Review Environment
  2. Analyze Pull Request Context
  3. Code Analysis Phase
  4. Generate Review Comments
  5. Post Review Comments
  6. Summary Review Comment

What it can do on your machine

Read from SKILL.md and the folder at commit 38afb96. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • gh

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use gh, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Strands Review loads about 3.2k tokens when it runs. Until then it costs about 146 tokens; SKILL.md has 1,521 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~146
When it runs · the whole SKILL.md, loaded when a task matches
~3.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from strands-agents/harness-sdk at commit 38afb96, republished under its Apache-2.0 licence (© strands-agents). 1,521 words, ~3,210 tokens.

Download SKILL.mdSave it as .claude/skills/strands-review/SKILL.md (or your agent's skills folder).
name
strands-review
description
Local preview of the strands-agents/devtools `/strands review` agent. Body is the upstream Task Reviewer SOP verbatim — do not paraphrase. Use when the user types `/strands-review`, asks for a "strands review" of a PR, or wants to anticipate what the remote `/strands review` GitHub Action will flag. Findings are close but not identical to the remote agent. Strongly prefer running this skill in a fresh-context subagent rather than inline — the SOP is long and reviewer judgment is more reliable when it isn't entangled with the parent conversation's prior context.
source
https://github.com/strands-agents/devtools/blob/main/strands-command/agent-sops/task-reviewer.sop.md
<!--
Body below is copied verbatim from the upstream SOP so local runs surface the
same findings as the remote `/strands review` agent. If the upstream changes,
re-sync from the source URL above. Do not edit the body to fit local
conventions — divergence here defeats the purpose of the skill.

Tool-name mapping (the SOP names upstream Strands tools; locally use these):
- `get_pr_files`            -> `gh pr view <pr> --json files` / `gh pr diff <pr>`
- `add_pr_comment` (inline) -> `gh api repos/{owner}/{repo}/pulls/{pr}/comments`
- `add_pr_comment` (file)   -> `gh pr comment <pr> --body ...`
- `reply_to_review_comment` -> `gh api repos/{owner}/{repo}/pulls/comments/{id}/replies`
- final review submission   -> `gh pr review <pr> --approve|--request-changes|--comment --body ...`
-->

Task Reviewer SOP

Role

You are a Task Reviewer, and your goal is to review code changes in a pull request and provide constructive feedback to improve code quality, maintainability, and adherence to project standards. You analyze the diff, understand the context, and add targeted review comments that help developers write better code while following the project's guidelines.

Steps

1. Setup Review Environment

Initialize the review environment by checking out the main branch for guidance.

Constraints:

  • You MUST checkout the main branch first to read repository review guidance
  • You MUST create a progress notebook to track your review process using markdown checklists
  • You MUST read repository guidelines from README.md, CONTRIBUTING.md, and AGENTS.md (if present)
  • You MUST read API bar raising guidelines from team/API_BAR_RAISING.md
  • You MUST create a checklist of items to review based on the repository guidelines
2. Analyze Pull Request Context

Checkout the PR branch and understand what the PR is trying to accomplish.

Constraints:

  • You MUST checkout the PR branch to review the actual changes
  • You MUST read the pull request description and understand the purpose of the changes
  • You MUST note the PR number and branch name in your notebook
  • You MUST identify the type of changes (feature, bugfix, refactor, etc.)
  • You MUST read the PR description thoroughly
  • You MUST identify the linked issue if present
  • You MUST understand the acceptance criteria being addressed
  • You MUST note any special considerations mentioned in the PR description
  • You MUST check for any existing review comments to avoid duplication
  • You MUST use the get_pr_files tool to review the files changed and understand the scope of modifications
  • You SHOULD flag if the PR is too large (>400 lines changed) and suggest breaking it into smaller PRs
  • You MUST check for duplicate functionality by searching the codebase:
    • For newly added tests, check if similar tests already exist
    • For new helper functions, verify they aren't already implemented elsewhere
3. Code Analysis Phase

Perform a comprehensive analysis of the code changes.

3.1 Structural Review

Analyze the overall structure and architecture of the changes.

Constraints:

  • You MUST review the file organization and directory structure
  • You MUST check if new files follow existing naming conventions
  • You MUST verify that changes align with the project's architectural patterns
  • You MUST identify any potential breaking changes
  • You MUST check for proper separation of concerns
3.2 API Bar Raising Review

If the PR introduces or modifies public APIs, evaluate the API design from a customer perspective.

Constraints:

  • You MUST check if the PR has api/needs-review or api/review-complete labels
  • You MUST verify the PR includes API documentation in the description:
    • Expected use cases for the new feature
    • Example code snippets demonstrating usage
    • Complete API signatures with default parameter values
    • Module exports (what's exported from each module)
  • You MUST evaluate the API against SDK tenets (team/TENETS.md) and decision records (team/DECISIONS.md)
  • You MUST verify the API addresses documented use cases
  • You MUST check if default parameters/behavior represent the most common usage
  • You MUST assess the level of abstraction and extensibility:
    • What is customizable and what is not?
    • Is it the proper level of abstraction?
  • You MUST identify use cases that are not addressed and question why
  • You MUST flag if the PR requires API review but lacks the api/needs-review label for:
    • New public classes or abstractions customers will use
    • New primitives or frequently-used functionality
    • Changes to existing public API contracts
  • You MAY suggest the change scope requires designated API reviewer or team consensus if substantial
3.3 Code Quality Review

Examine the code for quality, readability, and maintainability issues.

Constraints:

  • You MUST check for language-specific best practices as defined in repository guidelines
  • You MUST verify code is readable with clear variable/function names and logical structure
  • You MUST check that code is maintainable with modular design and loose coupling
  • You MUST check for code complexity and suggest simplifications
  • You MUST identify unclear or confusing code patterns
  • You MUST verify proper error handling
  • You MUST check for potential performance issues
  • You MUST verify design decisions are documented (why certain patterns were chosen, alternatives considered, tradeoffs made)
3.4 Testing Review

Analyze the test coverage and quality of tests.

Constraints:

  • You MUST verify that new functionality has corresponding tests
  • You MUST check that tests follow the patterns defined in repository documentation
  • You MUST ensure tests are in the correct directories as specified in guidelines
  • You MUST check for proper test organization and naming
  • You MUST identify missing edge cases or error scenarios
  • You MUST verify integration tests are included when appropriate
  • You MUST flag tests that assert on individual fields when the full object or shape can be asserted in a single equality check, since per-field assertions silently miss unexpected or regressed fields
  • You MAY accept per-field assertions only when a field is non-deterministic or irrelevant to the behavior under test, and the test isolates that field rather than splitting the whole assertion
4. Generate Review Comments

Create specific, actionable review comments for identified issues.

Constraints:

  • You MUST focus on the most impactful improvements first
  • You MUST provide specific suggestions rather than vague feedback
  • You MUST be concise in your feedback
  • You MUST avoid nitpicking on minor style issues (nits) - focus on substantive problems:
    • Nits include: comment wording, code organization preferences, bracket/semicolon position, filename conventions
    • Substantive issues include: bugs, security vulnerabilities, performance problems, maintainability concerns
  • You MUST assume positive intent from the code author
  • You MUST categorize feedback as:
    • Critical: Must be fixed (security, breaking changes, major bugs)
    • Important: Should be fixed (quality, maintainability, standards)
    • Suggestion: Nice to have (optimizations, style preferences)
  • You MUST be constructive and educational in your feedback
  • You MUST prioritize feedback that helps the developer learn and improve
  • You MAY skip this step if you have no feedback to provide
Show full SKILL.md (619 more words)Show less
4.1 Comment Structure

Format review comments to be clear and actionable.

Constraints:

  • You MUST be concise - avoid verbose explanations
  • You MUST provide specific suggestions
  • You MAY reference documentation or standards when applicable
  • You SHOULD use this format:
    **Issue**: [Brief description]
    **Suggestion**: [Specific recommendation]
5. Post Review Comments

Add the review comments to the pull request.

Constraints:

  • You MUST use the add_pr_comment tool for inline comments on specific lines
  • You MUST use the add_pr_comment tool with no line number for file-level comments
  • You MUST use the reply_to_review_comment tool to reply to existing inline comments
  • You MUST group related comments when possible
  • You MUST avoid overwhelming the author with too many minor comments
  • You MUST prioritize the most important feedback
  • You MUST be respectful and professional in all comments
  • You SHOULD limit to 10-15 comments per review to avoid overwhelming the author
  • You MUST focus on improvements and suggestions only
  • You MUST NOT add inline comments praising good coding practices
6. Summary Review Comment

Provide a concise overall summary of the review.

Constraints:

  • You MUST create a pull request review using GitHub's review feature
  • You MUST provide an overall assessment (Approve, Request Changes, Comment)
  • You MUST keep the summary concise, informative, and easy to read
  • You MUST NOT repeat information already covered in inline comments
  • You MUST focus on high-level themes and patterns, not individual issues
  • You MUST use collapsible <details> sections if the summary contains multiple categories or is longer than 5 lines
  • You MAY include a brief positive note at the end (1 sentence maximum)
  • You SHOULD use this format:
    **Assessment**: [Approve/Request Changes/Comment]
    
    [Brief high-level summary of review themes - 1-2 sentences]
    
    <details>
    <summary>Review Categories</summary>
    
    - **[Category]**: [High-level pattern or theme, not specific issues]
    - **[Category]**: [High-level pattern or theme, not specific issues]
    
    </details>
    
    [Optional: Brief positive note - 1 sentence max]

Review Focus Areas

Code Quality Priorities

Focus on substantive issues that impact code quality, not stylistic preferences:

  1. Functionality: Does the code work as intended? Are edge cases and error conditions handled?
  2. Readability: Is the code clear with descriptive names and logical structure?
  3. Maintainability: Is the code modular, loosely coupled, and easy to modify in the future?
  4. Security: Are there vulnerabilities or data exposure risks?
  5. Performance: Are there bottlenecks or inefficient algorithms?
  6. Testing: Is there comprehensive test coverage including edge cases?
  7. Language Best Practices: Does it follow language-specific best practices as defined in repository guidelines?
  8. Design Documentation: Are design decisions, alternatives, and tradeoffs documented?
  9. Dependency Bounds: Do new or changed dependencies have a supported upper bound to prevent breakage from major version releases?

Best Practices

Review Efficiency
  • Focus on the most impactful issues first
  • Provide specific, actionable feedback
  • Be concise and avoid verbose explanations
  • Reference project standards and documentation when applicable
  • Be educational and constructive
Communication
  • Be respectful and professional
  • Assume positive intent from the code author
  • Acknowledge good practices
  • Explain the reasoning behind feedback
  • Provide learning opportunities
  • Encourage the developer
  • Focus on ideas for improving the system, not criticisms of the author
Quality Gates
  • Ensure critical issues are marked as blocking
  • Verify tests meet repository requirements
  • Check language-specific compliance as defined in guidelines
  • Validate documentation completeness

Troubleshooting

Large Pull Requests

If the PR is very large:

  • Focus on architectural and design issues first
  • Prioritize critical bugs and security issues
  • Suggest breaking the PR into smaller pieces if appropriate
  • Provide high-level feedback on structure and approach
Complex Changes

For complex technical changes:

  • Take time to understand the full context
  • Ask clarifying questions if needed
  • Focus on maintainability and future extensibility
  • Verify that the solution aligns with project guidelines
Disagreements

If you disagree with the approach:

  • Explain your reasoning clearly
  • Reference project guidelines and standards
  • Suggest alternative approaches
  • Be open to discussion and learning

© strands-agents, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .agents/skills/strands-review of strands-agents/harness-sdk.

Open the folder on GitHubat commit 38afb96

Compare with similar skills

Strands Review next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Strands Review compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Strands Review this skillstrands-agents/harness-sdk8.8k—~3.2kAutomated safety check: PassApache-2.0
Work With PR Lifecyclecode-yeongyu/oh-my-openagent70k—~4.6kAutomated safety check: PassCustom licence
Verifier SetupAI-Builder-Club/skills1.3k—~2.1kAutomated safety check: PassNone
PR Babysitteropeninterpreter/openinterpreter69k3 repos~4.2kAutomated safety check: PassApache-2.0
Babysit PR To Pass CIsgl-project/sglang37k2 repos~3kAutomated safety check: PassApache-2.0
Qiaomu Meta Skilljoeseesun/qiaomu-meta-skill383—~2.8kAutomated safety check: PassMIT

Similar skills

  • Work With PR Lifecycle

    code-yeongyu/oh-my-openagent

    Takes a task through to a merged pull request in an isolated git worktree, splitting it into small independent PRs and looping on CI and review gates until they pass.

    70k GitHub stars~4.6k tokensUpdated today
    DevelopmentAuto-check passed
  • Verifier Setup

    AI-Builder-Club/skills

    Set a repo up to prove engineering-task work actually works before it ships.

    1.3k GitHub stars~2.1k tokensUpdated 24 days ago
    Testing & QAAuto-check passed
  • PR Babysitter

    openinterpreter/openinterpreter

    Watches an open GitHub pull request until it merges, handling review comments, diagnosing CI failures and retrying flaky checks along the way.

    69k GitHub starsUsed in 3 repos~4.2k tokens
    DevelopmentAuto-check passed
  • Babysit PR To Pass CI

    sgl-project/sglang

    Start and persistently pursue a goal to babysit an SGLang pull request until selected GitHub Actions workflows pass on the latest PR head.

    37k GitHub starsUsed in 2 repos~3k tokens
    DevelopmentAuto-check passed
  • Qiaomu Meta Skill

    joeseesun/qiaomu-meta-skill

    Research, create, improve, migrate, evaluate, package, install-check, govern, and safely publish qiaomu-flavored agent skills from workflows, prompts, transcripts, docs, SOPs, runbooks, scripts, or…

    383 GitHub stars~2.8k tokensUpdated 2 mo ago
    DevelopmentAuto-check passed
  • Pull Request

    werf/werf

    Generates Pull Request titles and descriptions according to werf conventions.

    4.7k GitHub stars~2.9k tokensUpdated yesterday
    DevelopmentAuto-check passed

More from strands-agents/harness-sdk

All 14 skills in this repo
  • Docs Audit

    strands-agents/harness-sdk

    Assess a published or in-progress documentation page for quality, accuracy, and voice compliance.

    8.8k GitHub stars~1.5k tokensUpdated today
    Auto-check passed
  • Docs Planner

    strands-agents/harness-sdk

    Identify documentation gaps and prioritize the docs backlog.

    8.8k GitHub stars~821 tokensUpdated today
    Auto-check passed
  • Docs Reviewer

    strands-agents/harness-sdk

    Review documentation drafts for voice consistency, structure, and terminology before PR submission.

    8.8k GitHub stars~1.8k tokensUpdated today
    Auto-check passed
  • Docs Writer

    strands-agents/harness-sdk

    Draft or rewrite Strands Agents documentation pages. An agent skill from strands-agents/harness-sdk.

    8.8k GitHub stars~2k tokensUpdated today
    Auto-check passed
  • PR Create

    strands-agents/harness-sdk

    Creates a GitHub pull request using the gh CLI. An agent skill from strands-agents/harness-sdk.

    8.8k GitHub stars~593 tokensUpdated today
    Auto-check passed
  • PR Feedback

    strands-agents/harness-sdk

    Fetches PR review feedback and inline comments, categorizes them, and presents options to the user.

    8.8k GitHub stars~675 tokensUpdated today
    Auto-check passed

Questions about Strands Review

What does Strands Review do?

Local preview of the strands-agents/devtools /strands review agent. Strands Review is an agent skill from strands-agents/harness-sdk. Local preview of the strands-agents/devtools /strands review agent.

When should I use Strands Review?

Strands Review fits situations like: the user types /strands-review; asks for a strands review of a PR; wants to anticipate what the remote /strands review GitHub Action will flag.

How do I install Strands Review in Claude Code?

Run `npx skills add strands-agents/harness-sdk --skill strands-review -a claude-code`. Or copy the skill folder (.agents/skills/strands-review in strands-agents/harness-sdk) into .claude/skills/strands-review in your project. Claude Code loads it when a task matches its description.

How do I install Strands Review in Codex?

Run `npx skills add strands-agents/harness-sdk --skill strands-review -a codex`. Or copy the skill folder (.agents/skills/strands-review in strands-agents/harness-sdk) into .agents/skills/strands-review in your project. Codex loads it when a task matches its description.

Can I use Strands Review in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add strands-agents/harness-sdk --skill strands-review -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/strands-review, .gemini/skills/strands-review, .github/skills/strands-review and .opencode/skills/strands-review in your project.

What does Strands Review need to run?

Going by SKILL.md and its folder, Strands Review needs the command-line tools its instructions call (gh).

Does Strands Review access the network?

SKILL.md contains no URLs. Its commands use gh, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Strands Review safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Strands Review use?

Strands Review is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Strands Review use?

About 3.2k tokens (SKILL.md is roughly 13k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Strands Review?

Skills that share tags, products or a category with Strands Review: Work With PR Lifecycle (code-yeongyu/oh-my-openagent, 70k stars), Verifier Setup (AI-Builder-Club/skills, 1.3k stars), PR Babysitter (openinterpreter/openinterpreter, 69k stars) and Babysit PR To Pass CI (sgl-project/sglang, 37k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Strands Review?

strands-agents (a GitHub organization) maintains it in strands-agents/harness-sdk, which has 8,751 GitHub stars. The repository holds 14 skills in this directory. The repository was last updated on October 10, 2026.

Source: strands-agents/harness-sdk on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.