code review frontend

Engineering Community

Skill notes

Guardrails

Review only the changed behavior and its direct integration points. Do not recommend broad rewrites, abstractions, or style churn unless they prevent a verified bug. If evidence is weak, label the assumption or omit the finding. Prefer findings with a concrete repro, failing test, config violation, or line-level reasoning.

Skill Directory

Sub-skill paths (read when relevant): $SKILL_DIR/checklist.md — base review checklist $SKILL_DIR/typescript-react-reviewer/SKILL.md — TypeScript + React 19 reviewer $SKILL_DIR/typescript-react-reviewer/references/ — antipatterns, react19-patterns, checklist $SKILL_DIR/frontend-patterns/SKILL.md — React component & state patterns

Step 0: Codebase Context Discovery

Before anything else, check whether the codebase has been mapped and its conventions are known. Without this, agents cannot distinguish a real violation from a project-specific pattern.

0a. Check .planning/ for existing docs

Read any files that describe: Architecture, conventions, coding standards Tech stack, frameworks in use CI/CD pipeline or quality enforcement If .planning/ has sufficient context → proceed to Step 0b.

0b. Discover Quality Tooling (Source of Truth)

Run these commands to find the project's actual quality standards. These become the ground truth that agents cite in findings. Read each config file found. Extract: Lint rules enforced (eslint rules/plugins active) Format rules (printWidth, semi, quotes, trailing commas)

Fluxly profile Runtime and permissions are inferred by Fluxly static checks, not author declarations or runtime test results.
Domain
Engineering
Author-declared agents
No explicit declaration found; this is not inferred or tested compatibility
Static check
88 / 100 · heuristic scan, not runtime safety proof
Runtime evidence
Not run; no isolated runtime test evidence is currently recorded
Author / version / license
@tomevault-io · no license declared
Token usage
Heavy
Setup complexity
Guided setup
External API key
Not required
Operating systems
macOS · Linux · Windows
Runtime requirements
No explicit requirements detected
Detected file/system behavior
  • Read-only
  • Write / modify
  • Shell exec
Network behavior
External requests
Install commands
None (reference only)

Heads up: Tool permissions are not declared; actual access depends on the runtime.

01
Step 0: Codebase Context Discovery

Before anything else, check whether the codebase has been mapped and its conventions are known. Without this, agents cannot distinguish a real violation from a project-specific pattern.

02
Step 1: Branch + Diff Check

If on base branch → "Nothing to review — on base branch." → stop. No diff → same message → stop. Read `.planning/tasks/TODOS.md` if it exists.

03
Step 2: Scope Detection

Based on files in the diff, decide which specialist agents to spawn: Files changed · Agent Any file (always) · `base-checklist` `.ts`, `.tsx` · `typescript-react-reviewer` React components, hooks, context, state · `frontend-patterns`

04
Step 3: Spawn Sub-Agents in Parallel

Spawn all applicable agents in one message (parallel execution). Pass each agent: The QUALITY_CONTEXT discovered in Step 0b The base branch name The skill files to read Agents do not fix code. They report findings only.

05
Step 4: Collect Reports

Wait for all agents. Collect all reports. Count totals by severity and evidence type.

06
Step 5: Synthesize — Main Agent Review

Read every agent report. Cross-check: if an agent cites `file:line`, verify that line exists in the diff. If it doesn't match → drop the finding and note it was unverifiable. De-duplicate: Same file:line flagged by multiple agents → merge into one finding, list all agents that caught it. Never introduce new findings here that weren't in an agent report. Your job is to

Discussion

Powered by GitHub Discussions. Sign in with GitHub to comment, react, or subscribe.