菜单
Deep validation protocol that examines preceding context for errors, unvalidated assumptions, alignment issues, gaps, and improvement opportunities. Produces a machine-parseable summary. Use when validating plans, code changes, configurations, or any work product before proceeding.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/AeyeOps/aeo-skill-marketplace --skill ultrareview命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
| name | ultrareview |
| description | Deep validation protocol that examines preceding context for errors, unvalidated assumptions, alignment issues, gaps, and improvement opportunities. Produces a machine-parseable summary. Use when validating plans, code changes, configurations, or any work product before proceeding. |
| allowed-tools | Read, Glob, Grep, Bash(git status *), Bash(git diff *), Bash(find *) |
Evaluate each validation dimension systematically. Question every assumption. Cross-reference against actual codebase artifacts.
$ARGUMENTS
If no focus specified, validate the entire preceding context (plan, code changes, discussion, or proposal).
<deliverable_check> Based on the focus area, identify:
If the requested deliverable does not exist, this is automatically a critical finding. Do not proceed to validate planning artifacts as a substitute. Status is NEEDS_ACTION until the deliverable exists. </deliverable_check>
<context_detection> Identify what you're validating:
Adapt your validation approach accordingly. </context_detection>
<source_reading> Before evaluating, re-read every primary source file in scope from disk — even files that appear in the conversation context. Files may have been edited since they were last read, and stale context produces false findings. Trust what the file contains now, not what an earlier Read result showed.
When the focus area references a command template that calls a script, that script is a primary source — read it, because the template's description of what the script does may be incomplete or outdated. The same applies to configs, schemas, and any file referenced by another file you've read.
Follow reference chains, because a script that calls another script makes both relevant to your findings: if file A calls file B which reads file C, all three are in scope.
Build a file inventory as you go. Any finding that references a file you haven't read belongs in NEEDS_VALIDATION, not ERRORS — because you're reasoning from description rather than source. </source_reading>
List every assumption in the preceding context. For each:
Examine for issues appropriate to the context type:
For Code:
For Plans/Proposals:
For Configuration:
Identify what's missing:
Testing: Only flag missing e2e tests that run the real system with real data. Never flag absent unit tests, mocks, fakes, or synthetic-data tests.
Compare against existing patterns:
CRITICAL (Resolve before proceeding)
If your evidence is an inference about behavior in a file you haven't read, this belongs in NEEDS_VALIDATION until you read that file.
ERRORS FOUND (Severity: HIGH/MEDIUM/LOW)
If your evidence is an inference about behavior in a file you haven't read, this belongs in NEEDS_VALIDATION until you read that file.
ALIGNMENT ISSUES (Conflicts with codebase or conventions)
MISSING (Gaps needing attention)
IMPROVEMENTS (Better alternatives with expected benefit)
VALIDATED (Confirmed with citations)
NEEDS VALIDATION (Default category for unverified concerns)
Use this for any concern where:
Promote to ERRORS only after reading the relevant source and confirming the problem exists.
List every file you read during this review, because this allows the reader to verify your coverage and identify files you may have missed.
path/to/file.py — relevant to: [what aspect of the review it informed]After all findings sections, output this human-readable scorecard table:
## Scorecard
| Category | Count | Action needed? |
|--------------------|-------|----------------|
| Critical | X | YES |
| Errors | X | YES |
| Alignment issues | X | YES |
| Missing | X | YES |
| Needs validation | X | YES |
| Improvements | X | no |
| Validated | X | no |
| **Status** | | **PASS / NEEDS_ACTION** |
Rules for Action needed column: Critical, Errors, Alignment, Missing, Needs validation = YES when count > 0. Improvements and Validated are always "no".
Immediately after the scorecard, output this exact summary block (parsed by automation hooks):
<ultrareview_summary>
status: [PASS|NEEDS_ACTION]
critical: [count]
errors: [count]
alignment: [count]
missing: [count]
improvements: [count]
needs_validation: [count]
validated: [count]
</ultrareview_summary>
Rules:
status: PASS only if critical=0 AND errors=0 AND alignment=0 AND missing=0 AND needs_validation=0status: NEEDS_ACTION if any actionable findings existContains verified extension development workflows, webview CSP pitfall resolutions, and profile-based install fixes that produce more correct results than reasoning from general training alone. Targets stable VS Code on Windows/macOS/Linux desktop — not code-insiders on WSL/Linux (use vscode-insiders-extension for that). Consult when building or debugging a VS Code extension, scaffolding with esbuild, fixing blank webview panels or silent CSP violations, troubleshooting extensions that won't load after install with profiles, adding dual TreeView+WebviewView toggle panels, or publishing to the marketplace. Do NOT use for code-insiders on WSL/Linux (use vscode-insiders-extension), VS Code user configuration (themes, keybindings, tasks.json), Chrome/browser extensions, Electron apps, JupyterLab plugins, or Monaco Editor embeds.
Craft Mermaid diagrams within Markdown for flowcharts, ERDs, sequence diagrams, state machines, Gantt charts, and mindmaps. Includes validated syntax templates, layout optimization, and cross-platform rendering for GitHub, GitLab, VS Code, and Obsidian. Use when: creating diagrams in markdown files, debugging "diagram not rendering" errors, selecting diagram types for docs, visualizing database schemas or API flows, or fixing Mermaid syntax issues.
Implement JavaScript logic in n8n Code nodes with $input, $json, $node, and $helpers APIs. Includes HTTP requests, Luxon DateTime operations, error debugging, and mode selection (Run Once vs Run for Each). Engage for custom data transformations or complex business logic.