Deepen

Deepen Plan

Take an existing plan (from /cplan) and enhance each section with parallel research agents. Each major element gets its own dedicated research sub-agent to find:

• Best practices and industry patterns
• Performance optimizations
• UI/UX improvements (if applicable)
• Quality enhancements and edge cases
• Real-world implementation examples

The result is a deeply grounded, production-ready plan with concrete implementation details.

Subagents

This skill dispatches these research subagents in parallel:

• cexplore — codebase patterns and conventions
• cdocs — external documentation via Context7 MCP
• clearnings — past solutions from docs/solutions/
• cbestpractices — industry standards and community patterns

It also launches all available *-reviewer agents against the plan content.

Workflow

Step 0: Locate the Plan

1. If a plan path was provided as argument, use it
2. Otherwise, read the path from docs/plans/.latest
3. If neither exists, list docs/plans/ and ask the user which plan to deepen

Do not proceed until you have a valid plan file.

Step 1: Parse and Analyze Plan Structure

Read the plan file and extract:

• Overview/Problem Statement
• Proposed Solution sections
• Technical Approach/Architecture
• Implementation phases/steps
• Code examples and file references
• Acceptance criteria
• Any UI/UX components mentioned
• Technologies/frameworks mentioned

Create a section manifest:

Section 1: [Title] - [Brief description of what to research]
Section 2: [Title] - [Brief description of what to research]
...

Step 2: Research (Parallel)

Launch research sub-agents in parallel for each major section:

For each section, spawn a cexplore agent:

• Research codebase patterns and conventions for the section topic
• Find existing implementations and architecture decisions

Spawn cbestpractices for each major technology or pattern:

• Research industry standards and community conventions
• Find recommended patterns, anti-patterns, and production guidance
• Identify performance considerations and common pitfalls

For each technology/framework mentioned, spawn a cdocs agent:

• Fetch current documentation via Context7 MCP
• Find version-specific patterns and constraints
• Get concrete code examples

Spawn clearnings for documented solutions:

• Check docs/solutions/ for relevant past solutions
• Surface institutional knowledge that applies to this plan

Step 3: Discover and Run Review Agents

Launch ALL available review agents in parallel against the plan content. Do not filter by relevance — let each reviewer decide what applies. The goal is maximum coverage.

Step 4: Synthesize Findings

Collect outputs from all agents and:

• Extract concrete recommendations (actionable items)
• Extract code patterns and examples
• Identify anti-patterns to avoid
• Note performance and security considerations
• Surface edge cases
• Deduplicate and prioritize by impact
• Flag conflicting advice for human review
• Group findings by plan section

Step 5: Enhance Plan Sections

For each section, add research insights below the original content:

## [Original Section Title]

[Original content preserved]

### Research Insights

**Best Practices:**
- [Concrete recommendation]

**Performance Considerations:**
- [Optimization opportunity]

**Implementation Details:**
```[language]
// Concrete code example from research

Edge Cases:

• [Edge case and how to handle]

References:

• [Documentation URL]

### Step 6: Add Enhancement Summary

At the top of the plan, add:

```markdown
## Enhancement Summary

**Deepened on:** [Date]
**Sections enhanced:** [Count]
**Research agents used:** [List]

### Key Improvements
1. [Major improvement 1]
2. [Major improvement 2]
3. [Major improvement 3]

### New Considerations Discovered
- [Important finding 1]
- [Important finding 2]

Step 7: Handover

1. Update the plan file in place
2. Present a brief summary of what was added

Use #askQuestions to ask what the user wants to do next:

After the user picks a next skill, announce the handover and load the chosen skill.

Response Rules

• Never echo full file contents into chat — reference by path
• Keep chat responses under 500 words
• Preserve ALL original plan content — only add, never remove
• Mark all additions clearly as "Research Insights"

Guidelines

• Never write code in the plan — only research and enhance
• Include specific file paths and documentation URLs as references
• Make recommendations concrete and actionable
• When agents disagree, present both perspectives and flag for human decision