Code Review Instructions

Review Checklist

• Unit and e2e tests are covering the changes well.
• Important: Changes follow the best practices documented in the relevant AGENTS.md files (read the ones that apply to the changed files):
  • e2e_playwright/AGENTS.md — for e2e tests (inside e2e_playwright/)
  • frontend/AGENTS.md — for frontend changes and unit tests (inside frontend/)
  • lib/tests/AGENTS.md — for Python unit tests (inside lib/tests/)
  • lib/AGENTS.md — for any Python changes (*.py files)
  • lib/streamlit/AGENTS.md — for any Python library changes (inside lib/streamlit/)
  • lib/streamlit/.agents/skills/AGENTS.md — for bundled agent skills (inside lib/streamlit/.agents/skills/)
  • proto/streamlit/proto/AGENTS.md — for protobuf changes (inside proto/streamlit/proto/)
• Product alignment is explicitly assessed for user-facing changes. Treat a change as user-facing when it adds or modifies a public API, configuration option, CLI surface, rendered UI or interaction, default, error message, or other externally observable behavior.
  • Read specs/AGENTS.md and evaluate the change against its "Principles of Streamlit API Design."
  • Search specs/ for a relevant product spec. Product decisions explicitly described in a product spec that is already merged into the PR's base branch are considered approved and aligned; do not relitigate them. Instead, verify that the implementation follows the spec and separately assess any user-facing behavior that the merged spec does not cover. A new or modified spec in the current PR is not already approved by this rule.
  • A new user-facing feature or significant public API change requires a product spec already merged into the base branch. If none exists, call this out as a merge-blocking Product Alignment issue: merge the product spec first, then the implementation (see specs/README.md and specs/AGENTS.md). A spec that exists only on this implementation PR is not sufficient. Bug fixes, DevOps work, and small non-controversial enhancements do not need a spec.
  • If no merged spec covers a product decision, that absence is not approval. Request changes for material product/API alignment issues.
  • Inspect analogous Streamlit APIs, configs, and behaviors. Check that the proposed surface uses established names, defaults, interaction patterns, return types, and error behavior. For public APIs, follow the "Docstrings for Public API" best practices in lib/streamlit/AGENTS.md and compare docstrings with analogous API methods and functions for consistent terminology, parameter descriptions, documented behavior, and examples.
  • Confirm that the change solves a clear user problem, keeps the common case simple, exposes complexity progressively, composes with existing features, and adds no more permanent user-facing surface area than its value justifies.
  • Consider the complete user experience, including discoverability, helpful failures, backwards-compatible evolution, and behavior across supported platforms when relevant.
  • If the PR has no user-facing product impact, state that explicitly rather than inventing product concerns.
• Assess whether the implementation's complexity is proportionate to the value it provides. Call out unnecessary abstractions, indirection, state, dependencies, or maintenance burden.
• Assess performance impact. For backend or frontend runtime changes, apply the lib/streamlit/AGENTS.md ("Streamlit Backend Performance Hot Paths") and frontend/AGENTS.md ("Streamlit Frontend Performance Hot Paths") guidance. Watch for unnecessary renders or remounts, unstable props or callbacks, repeated parsing, copying, serialization, or full-data scans, blocking work, and inefficient algorithms. Hot-path changes should include performance coverage when practical.
• No risky aspects that could cause security issues or regressions. Pay closer attention to changes in these security-sensitive areas:
  • WebSocket connection handling, server endpoints, authentication, and session management
  • File upload, file/asset serving, and path traversal risks
  • Cookies, XSRF protection, CORS, cross-origin behavior, and security headers (CSP, etc.)
  • New backend or frontend dependencies, or requests to external assets/services
  • Runtime JavaScript execution (e.g., eval, unsafe-eval, Function() constructor)
  • Command/code injection risks (e.g., subprocess, exec, eval in Python)
  • HTML/Markdown rendering and sanitization (XSS risks)
  • iframe embedding and postMessage handling
  • Sensitive data handling (secrets, credentials, tokens)
  • st.login()/st.logout() and OAuth token handling
• External-test risk is explicitly assessed using /assessing-external-test-risk, and the review includes a clear external_test recommendation.
• Frontend changes follow accessibility best practices.
• The code follows other best practices from the Streamlit code base.

Instructions

1. Read the root AGENTS.md first to get an overview of the project.
2. Gather relevant context (branch diff, PR details if available).
3. Read and analyze the changed files to understand the full context.
4. Important: Read the relevant sub-directory AGENTS.md files based on changed files (see checklist above).
5. Classify whether the PR has user-facing product impact. If it does, read specs/AGENTS.md, inspect comparable existing Streamlit surfaces, and perform the product alignment assessment from the checklist. If it does not, mark the Product Alignment section as not applicable and explain why briefly.
6. Assess performance impact, including the backend and frontend hot paths.
7. Run an explicit external-test risk assessment using /assessing-external-test-risk and determine whether this branch should include @pytest.mark.external_test coverage.
8. Evaluate readability: run the /reviewing-readability skill on the changed code (comments, docstrings, naming) and the /reviewing-pr-description skill on the PR title/description if a PR exists, and include their findings and proposed rewrites in your review.
9. Perform a thorough code review based on the checklist above.
10. Write your review following the output format below.

Output Format

Write your review using valid GitHub Flavored Markdown in the following structure:

## Summary

[Brief overview and the main changes introduced.]

## Product Alignment

[State whether the PR has user-facing product impact. If yes, identify any relevant product
spec already merged into the base branch, treat its documented product decisions as approved,
and assess whether the implementation follows it. If this is a new user-facing feature or
significant public API change with no merged product spec, request that the spec be merged
before this implementation PR. For user-facing aspects not covered by an approved spec,
assess the user value and complexity tradeoff, consistency with analogous Streamlit surfaces,
and alignment with the principles in `specs/AGENTS.md`; clearly identify material issues that
warrant requested changes. If no, state why this section is not applicable.]

## Code Quality

[Brief assessment of code structure, patterns, and maintainability. Call out aspects whose
complexity appears disproportionate to the value they provide. Note any issues with specific
file references and line numbers.]

## Performance

[Assess runtime and frontend performance impact. Apply the backend and frontend hot-path
guidance. Call out unnecessary renders, remounts, repeated work, blocking operations, or
other regressions. Note coverage or benchmarks for hot-path changes. If there is no
meaningful performance impact, say why briefly.]

## Test Coverage

[Evaluation of unit and e2e test coverage. Are the changes adequately tested?]

## Backwards Compatibility

[Analysis of any breaking changes. Will this affect existing users?]

## Security & Risk

[Any security concerns or regression risks identified.]

## External test recommendation

[State `external_test` recommendation (Yes/No), triggered categories (or "None"), key evidence from changed files, suggested external test focus areas, and confidence plus assumptions/gaps.]

## Accessibility

[Assessment of accessibility considerations for frontend changes.]

## Readability

[Findings from the `/reviewing-readability` skill (comments, docstrings, naming) and, if a PR exists, the `/reviewing-pr-description` skill (title/description). Group by file; give the location, the issue, and the proposed rewrite.]

## Recommendations

[Specific suggestions for improvement, if any. Use a numbered list for actionable items.]

## Verdict

**[APPROVED / CHANGES REQUESTED]**: [One sentence summary of the overall assessment.]

Verdict criteria:
- **APPROVED**: If there are no critical/merge-blocking issues. Minor suggestions or optional improvements should not block approval — those can be addressed in follow-up PRs.
- **CHANGES REQUESTED**: Only use this for merge-blocking issues such as: bugs, security vulnerabilities, material performance regressions, breaking changes, missing required tests, a new user-facing feature or significant public API change without a product spec already merged into the base branch, or clear material violations of documented Streamlit product/API principles and patterns. Optional improvements, style preferences, and "nice to have" suggestions should NOT result in CHANGES REQUESTED.

---
*This is an automated AI review. Please verify the feedback and use your judgment.*

Important Notes

• Do NOT run linting, tests, or build commands - focus only on code review.
• Do NOT attempt to post comments, edit PRs, or perform any write operations.
• Focus on the root cause of issues, not cascading failures.
• Be specific with file references and line numbers when noting issues.
• Product feedback must cite concrete changed behavior and the relevant documented principle or established Streamlit pattern. A new user-facing feature or significant public API change without a product spec already merged into the base branch is merge-blocking: request that the spec land first. For smaller changes that do not require a spec, a missing merged spec is expected; still request changes for material alignment issues. Product questions and optional refinements are non-blocking; request changes when a mismatch would create material user harm or lasting, unjustified complexity in the public surface.
• Findings that are covered by inline comments should NOT be repeated in the PR-level review body. The PR-level review covers high-level and cross-cutting concerns only. Inline comments handle line-specific findings.