dd-docreview-grilling

Conducts interactive document reviews with diagram-based answers and structured TODO tracking.

Updated Jun 11, 2026
One-click install
npx skills add https://github.com/marcocpt/trae_skills --skill dd-docreview-grilling-marcocpt
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: dd-docreview-grilling
Source: https://github.com/marcocpt/trae_skills/tree/main/dd-docreview-grilling
Command: npx skills add https://github.com/marcocpt/trae_skills --skill dd-docreview-grilling-marcocpt

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reviewing long documents or verifying that code matches documentation is tedious and error-prone when done ad hoc. This Skill turns document review into a structured, interactive Q&A session where every answer is visualized, every finding is recorded as a trackable TODO, and confirmed issues are routed into batched fix workflows. ## Core Features & Use Cases - Two review modes: pure document review (design, completeness, ambiguity) or document-code consistency review that verifies claims against the actual codebase via search subagents. - Visual answers with anchors: every answer includes Mermaid diagrams (flowchart, state, sequence, class) or an HTML server for complex architectures, plus clickable file/line anchors. - Dual-track recording: findings are saved as Logseq-format TODO items or LATER index entries, written incrementally so nothing is lost if the session ends. - Batched remediation: TODOs are classified as bug vs feature, grouped by module into batches of up to 5, routed to bug-fix or feature-development workflows, and each batch produces a verification summary document with a fix commit SHA. - Use Case: You hand the assistant a design spec and ask it to verify the codebase section by section; each confirmed mismatch becomes a prioritized TODO, and at the end the issues are fixed in parallel-safe batches with per-batch verification steps. ## Quick Start Ask the assistant to review a document interactively, for example: "Review the doc at docs/design/auth-spec.md against the codebase, section by section."

Frequently Asked Questions about dd-docreview-grilling

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a document interactively with an AI assistant?

Provide the document path and trigger the review with a phrase like "review doc" or "grill this doc". The assistant reads the file, asks which section to start from, then answers each question with diagrams and asks whether to record findings as TODO or LATER items.

How to verify code matches documentation automatically?

Use the document-code consistency mode, which dispatches search subagents to inspect the codebase before answering any conformance question. Verified answers include clickable code location anchors, and confirmed mismatches become TODO items with a recommended fix file.

What format are the review TODO items saved in?

Findings are saved as Markdown task list items with Logseq-style key-value properties, including document location, current state, improvement suggestion, priority, and feature tags. The file path is confirmed at session start and items are appended immediately when marked.

When should I not use interactive document review?

Skip it for one-off questions that can be answered directly, for generating new documents from scratch, or for pure code review without a reference document. Those cases are better served by writing or code-review specific workflows.

How are confirmed review findings fixed in batches?

Each TODO is classified as bug or feature, grouped by functional module into batches of at most five, then routed to the matching fix workflow. Each batch updates the TODO status with the fix commit SHA and produces a verification summary with step-by-step manual checks.