deep-review
Scannednpx machina-cli add skill rana/yogananda-skills/deep-review --openclawRead all project markdown documents to ground in the project's actual state.
Deep Multi-Dimensional Review
This is the comprehensive quality gate. Check every dimension:
Coherence
- Cross-document alignment, identifier consistency, stated vs actual state
- Terminology consistency, bidirectional reference integrity
Gaps
- Missing decisions, unaddressed scenarios, dependency gaps
- Persona blind spots, phase transition gaps
Errors
- Incorrect cross-references, stale information, broken assumptions
- Wrong labels, associations, or identifiers
- Diagram endpoints still valid? Identifier sequences correct?
Architecture
- Pattern compliance, drift from stated principles
- Emerging patterns that should be codified
- Decisions that should be revisited given current state
Opportunities
- Simplifications available? Capabilities underexploited?
- Strategic gaps? Phase transitions that could be accelerated?
Focus area: $ARGUMENTS
Be proactive — propose concrete actions, not just observations. For every insight:
- What specific change would this imply? (code, design, decision, or document)
- Where exactly does it belong? (file, section, identifier)
- What's the next step to act on it?
Present as a prioritized action list. No changes to files — document only.
Output Management
Hard constraints:
- Segment output into groups of up to 10 findings, ordered by severity across all dimensions.
- If no $ARGUMENTS focus area is given, evaluate Coherence and Errors first (the most concrete dimensions), then Gaps if room remains.
- Write each segment incrementally. Do not accumulate a single large response.
- After completing each segment, continue immediately to the next. Do not wait for user input.
- Continue until ALL findings across all dimensions are reported. State the total count when complete.
- If the analysis surface is too large to complete in one session, state what was covered and what remains.
Document reading strategy:
- Read project documentation index/overview sections first. Only drill into specific sections where cross-references or identifiers need verification.
- Do not read the entire codebase — sample based on the focus area.
What questions would I benefit from asking?
What am I not asking?
You have complete design autonomy.
Source
git clone https://github.com/rana/yogananda-skills/blob/main/skills/deep-review/SKILL.mdView on GitHub Overview
Deep-review is a comprehensive quality gate that evaluates coherence, gaps, errors, architecture, and opportunities across project Markdown documents. It grounds decisions in the project’s actual state and surfaces concrete, prioritized actions for use before starting implementation or at phase boundaries.
How This Skill Works
The tool reads project Markdown documents to verify cross-document alignment, identifier consistency, and reference integrity. It then surfaces findings by dimension (Coherence, Gaps, Errors, Architecture, Opportunities) and returns a prioritized action list detailing the specific changes, exact location, and the next steps to act on.
When to Use It
- Before starting implementation of a new feature to establish a trusted baseline.
- At phase boundaries to validate alignment before handoffs or transitions.
- After major documentation updates to catch regressions and inconsistencies.
- During onboarding to align new team members with the current state.
- When a focus area (e.g., ARGUMENTS) is provided to tailor the review to that domain.
Quick Start
- Step 1: Run a deep-review pass over the project’s Markdown docs to surface findings across dimensions.
- Step 2: Inspect results with a focus on Coherence and Errors first, then evaluate Gaps and Architecture as time allows.
- Step 3: Produce a prioritized action list with concrete changes, target locations, and explicit next steps.
Best Practices
- Start with the project documentation index and overview sections to orient findings.
- Evaluate Coherence and Errors first, then Gaps if there is remaining capacity.
- Prioritize findings by severity and cross-dimension impact to drive action.
- For every insight, specify the exact change, the document or section it belongs to, and the next steps.
- Document actions as a backlog and re-run the review after applying changes to validate alignment.
Example Use Cases
- Pre-implementation quality gate for a data-platform migration to ensure cross-document references remain valid.
- Cross-document coherence check for a new feature set to align terminology and identifiers.
- Phase-transition review during a microservices refactor to surface decisions that should be codified.
- Detection of stale references and broken relationships after diagram updates.
- Revisions to phase-transition decisions in a product roadmap to accelerate critical handoffs.
