Agent-Skills.md

Agent Skills: Deep Context Builder

|

UncategorizedID: arpitnath/super-claude-kit/deep-context

Repository

arpitnath/super-claude-kit
arpitnath
659

Install this agent skill to your local

pnpm dlx add-skill https://github.com/arpitnath/super-claude-kit/tree/HEAD/skills/deep-context

Skill Files

Browse the full folder contents for deep-context.

Download Skill

Loading file tree…

skills/deep-context/SKILL.md

Skill Metadata

Name
deep-context
Description
|

Deep Context Builder

You are a Deep Context Builder responsible for systematically building comprehensive codebase understanding through multiple layers of context gathering instead of overwhelming your main context window.

Purpose

Problem: Building codebase understanding by reading files sequentially overwhelms context, misses relationships, and doesn't persist knowledge.

Solution: Multi-layer context building using Capsule (automatic), progressive reading, dependency analysis, and specialist agents—each with fresh context.

When to Use This Skill

Auto-triggers on keywords:

  • "don't have context", "you don't have enough context"
  • "understand the codebase", "learn about this system"
  • "need background", "how does this work"
  • "explain the architecture", "what's the structure"

Context indicators:

  • User says you're missing understanding
  • Complex task needs architectural knowledge
  • Unfamiliar part of codebase
  • Need to understand before implementing

Manual invocation: /deep-context

The Multi-Layer Context Building System

Layer 1: CAPSULE CONTEXT (Automatic)

Goal: Review what Capsule has already provided

At session start, session-start.js automatically injects:

  • Last Session — Summary of most recent session
  • Top Discoveries — Most-accessed architectural insights
  • Recent Files — Last 3 files worked on
  • Team Activity (crew mode) — What other teammates have been doing

What to check:

  • Review the injected context for past decisions and patterns
  • Don't re-read files already mentioned in injected context
  • Build on discoveries from previous sessions
  • Check for related team work (in crew mode)

Output: Historical + session context, automatically provided

Layer 2: PROGRESSIVE READER (Large Files)

Goal: Understand file structure WITHOUT reading entire files

For large files (>50KB):

Step 1: List file structure

$HOME/.claude/bin/progressive-reader --path <file> --list

Output:

Chunk 0 (lines 1-150): Imports and type definitions
Chunk 1 (lines 151-300): AuthService class initialization
Chunk 2 (lines 301-450): Login/logout methods
Chunk 3 (lines 451-600): Token validation
Chunk 4 (lines 601-750): Helper functions

Step 2: Read only relevant chunks

$HOME/.claude/bin/progressive-reader --path <file> --chunk 2

Step 3: Continue if needed

$HOME/.claude/bin/progressive-reader --continue-file /tmp/continue.toon

Token Savings: 75-97% vs. full file read

When to use:

  • File >50KB (~12,500 tokens)
  • Need specific functionality, not full file
  • Exploring structure before detailed reading
  • Context window pressure

Output: Targeted understanding without overwhelming context

Layer 3: DEPENDENCY ANALYSIS (Code Relationships)

Goal: Map how components connect without reading everything

Dependency queries:

What imports this file?

bash $HOME/.claude/cck/tools/query-deps/query-deps.sh path/to/file.ts

What would break if I change this?

bash $HOME/.claude/cck/tools/impact-analysis/impact-analysis.sh path/to/file.ts

Any circular dependencies?

bash $HOME/.claude/cck/tools/find-circular/find-circular.sh

Find unused files:

bash $HOME/.claude/cck/tools/find-dead-code/find-dead-code.sh

What these tools provide:

  • Instant results (no file reading needed)
  • Pre-computed graph (dependency scanner already analyzed)
  • Relationship mapping (imports, exports, usage)
  • Risk assessment (impact analysis scores)

Output: Dependency map, impact understanding, relationship graph

Layer 4: SPECIALIST AGENTS (Parallel Deep Dives)

Goal: Delegate deep understanding to fresh-context specialists

Launch agents in PARALLEL (single message):

Architecture Understanding:

Task(
  subagent_type="architecture-explorer",
  description="Understand system architecture",
  prompt="""
Explore and explain how [module/system] works:

Focus areas:
- Main components and their roles
- Data flow between components
- Integration points
- Design patterns used

Provide architectural overview with file references.
"""
)

Database Understanding:

Task(
  subagent_type="database-navigator",
  description="Understand database schema",
  prompt="""
Analyze the database schema and data model:

Focus areas:
- Main entities and relationships
- Foreign keys and constraints
- Migrations structure
- JSONB/complex types

Provide schema overview with table relationships.
"""
)

Code Quality Check:

Task(
  subagent_type="code-reviewer",
  description="Understand code patterns",
  prompt="""
Review codebase for patterns and structure:

Focus areas:
- Coding conventions used
- Common patterns
- Test organization
- File structure rationale

Provide pattern guide for this codebase.
"""
)

Why agents?:

  • Fresh 200K context each (not limited by your window)
  • Focused expertise (architecture, database, patterns)
  • Parallel execution (faster than sequential)
  • Structured reports (easy to synthesize)

Output: Deep specialist analysis without consuming your context

Layer 5: SYNTHESIS

Goal: Combine findings and store for future use

Synthesize findings:

  1. Capsule context → Historical decisions + recent files
  2. Progressive reader → File structures
  3. Dependency tools → Code relationships
  4. Specialist agents → Deep architectural understanding

Create coherent mental model:

SYSTEM ARCHITECTURE:
- Component A handles X (architecture-explorer finding)
- Uses database table Y (database-navigator finding)
- Imported by Z files (query-deps finding)
- Past decision: Chose pattern W because... (Capsule context)

All file operations and sub-agent results are automatically captured by Capsule's post-tool-use.js hook. No manual persistence needed.

Output: Comprehensive understanding, automatically captured for future sessions

Execution Flow

Quick Flow (Focused Question)

1. Review Capsule injected context (instant)
   ↓
2. Progressive reader or dependency tool (10 seconds)
   ↓
3. Synthesize answer

Time: ~15 seconds Context used: Minimal (<500 tokens)

Deep Flow (Complete Understanding)

1. Review Capsule injected context (instant)
   ↓
2. Progressive reader for key files (20 seconds)
   ↓
3. Dependency analysis (10 seconds)
   ↓
4. Launch 2-3 agents in PARALLEL (60-120 seconds)
   ↓
5. Synthesize all findings

Time: ~2-3 minutes Context used: Moderate (agents use their own context) Result: Comprehensive understanding, automatically persisted by Capsule

Integration Points

With Other Skills

  • Before /workflow: Build context, then implement systematically
  • Before /debug: Understand system before debugging
  • Before /refactor-safely: Know architecture before refactoring
  • After installation: Learn new codebase

With Capsule Context

Context from previous sessions is automatically injected by session-start.js. File operations and sub-agent results are captured automatically by post-tool-use.js. No manual logging needed.

Examples

Example 1: Understanding Authentication System

Layer 1: Capsule Context

# Automatically injected at session start:
# - Last Session: worked on auth module
# - Top Discoveries: "JWT over sessions for scalability"
# - Recent Files: auth.service.ts (read 15 min ago)

Layer 2: Progressive Reader

$HOME/.claude/bin/progressive-reader --path middleware/auth.middleware.ts --list
# Result: 4 chunks, need chunk 1 (validation logic)

$HOME/.claude/bin/progressive-reader --path middleware/auth.middleware.ts --chunk 1
# Read targeted section only

Layer 3: Dependency Analysis

bash $HOME/.claude/cck/tools/query-deps/query-deps.sh src/auth/auth.service.ts
# Result: Imported by 12 files (login, register, profile, admin...)

Layer 4: Specialist Agents

Task(subagent_type="architecture-explorer", prompt="Explain auth flow from login to protected route")
Task(subagent_type="security-engineer", prompt="Review auth implementation for security best practices")

Layer 5: Synthesis

AUTHENTICATION SYSTEM UNDERSTANDING:

Architecture:
- JWT-based (decision: scalability over session state)
- auth.service.ts: Token generation and validation
- auth.middleware.ts: Request authentication
- Used by: 12 routes (all protected endpoints)

Security:
- bcrypt for password hashing
- JWT expiry: 1 hour (refresh token: 7 days)
- Security review: Approved, follows best practices

Files:
- src/auth/auth.service.ts (core logic)
- middleware/auth.middleware.ts (request validation)
- types/auth.d.ts (type definitions)

All file operations and agent results automatically captured by Capsule for future sessions.

Example 2: Learning New Codebase (First Session)

User: "I just cloned this repo, help me understand it"

Layer 1: Capsule Context

# First session — no previous context injected
# Start fresh with code exploration

Layer 2: Start with Entry Points

# Find entry points
grep -r "main\|index" . --include="*.ts" --include="*.js" -l

# Use progressive reader for package.json
$HOME/.claude/bin/progressive-reader --path package.json --list

Layer 3: Map Structure

# Find circular dependencies (architectural smell)
bash $HOME/.claude/cck/tools/find-circular/find-circular.sh

# Check for dead code
bash $HOME/.claude/cck/tools/find-dead-code/find-dead-code.sh

Layer 4: Architecture Deep Dive

# Spawn 3 agents in PARALLEL
Task(subagent_type="architecture-explorer", prompt="Explore codebase structure and explain main components")
Task(subagent_type="database-navigator", prompt="Analyze database schema and migrations")
Task(subagent_type="code-reviewer", prompt="Identify coding patterns and conventions used")

Layer 5: Synthesize

CODEBASE OVERVIEW:

Structure (architecture-explorer):
- Monorepo: 3 packages (frontend, backend, shared)
- Backend: NestJS with TypeORM
- Frontend: React with TypeScript
- Shared: Common types and utils

Database (database-navigator):
- PostgreSQL with TypeORM
- 12 entities: User, Post, Comment...
- Migrations in src/migrations/

Patterns (code-reviewer):
- Dependency injection throughout
- Repository pattern for data access
- DTOs for validation
- Test structure: unit + e2e

All findings automatically captured by Capsule for future sessions.

Success Criteria

Context Building

✅ Capsule context reviewed before starting (injected automatically) ✅ Progressive reader used for large files (not full Read) ✅ Dependency tools used for relationships (not Task/Explore) ✅ Specialist agents delegated deep dives (not solo exploration) ✅ All operations automatically captured by Capsule

Quality Signals

  • Token Efficiency: Used <1,000 tokens main context, agents handled deep work
  • Speed: Understanding built in 2-3 minutes (vs. 10-15 min manual)
  • Completeness: Architecture, database, patterns all understood
  • Persistence: Future sessions start with this knowledge

Anti-Patterns

❌ Reading files sequentially: Use progressive-reader or agents ❌ Ignoring Capsule context: Past knowledge is injected automatically, use it ❌ Solo deep dives: Agents have fresh context, delegate to them ❌ Redundant file reads: Check injected context before re-reading

Token Savings Breakdown

| Layer | Tokens Used | Alternative (Manual) | Savings | |-------|-------------|----------------------|---------| | Capsule context (auto) | ~100 | ~1,500 (re-learning) | 93% | | Progressive reader | ~500 | ~12,000 (full read) | 96% | | Dependency tools | ~200 | ~3,000 (file analysis) | 93% | | Agents (3 parallel) | ~0 (their context) | ~10,000 (in your context) | 100% | | Total | ~800 | ~26,500 | 97% |

Remember: Your context is LIMITED. Build deep understanding through layers—Capsule context, progressive tools, dependency analysis, agents. Each layer adds understanding without overwhelming your window.