Research Methodology

Research Methodology Skill

Purpose

Ensure that code is informed by accurate, up-to-date knowledge — not stale training data assumptions. This skill provides a structured process for researching technologies, APIs, and patterns before writing code.

When to Invoke

• Before implementing features using unfamiliar technologies
• When evaluating library or framework options
• When the Architect needs to inform design decisions with current best practices
• When any builder agent encounters an API or pattern they haven't used before

Research Process

1. Topic Decomposition

Break the research need into specific, searchable topics (2-5 keywords each):

Example — "Task CRUD API with Supabase":
  Topics:
  1. Supabase client library JavaScript/TypeScript
  2. Supabase Row Level Security policies
  3. Supabase real-time subscriptions
  4. PostgreSQL UUID primary key patterns
  5. TypeScript type generation from Supabase

2. Multi-Tool Search Strategy

Search for each topic using the best available tool, in priority order:

Search strategy:

# Start broad, then narrow:
search("Supabase RLS policies")           # Broad — find the right doc page
read_page("https://supabase.com/docs/...") # Deep — read the specific page

# Search for gotchas and edge cases:
search("Supabase RLS common mistakes")
search("Supabase RLS performance implications")

3. Document Findings

Create a research log for each feature or investigation:

Path: docs/research_logs/{feature_name}.md

Naming convention:

docs/research_logs/
├── epic1_auth.md           # Epic 1: Authentication
├── epic2_task_crud.md       # Epic 2: Task CRUD
├── supabase_rls_evaluation.md  # Technology evaluation
└── {feature_name}.md       # Pattern: one log per feature

Research log template:

# Research: {Feature or Topic Name}
Date: {date}
Researcher: {agent name}

## Topics Investigated
1. {topic} — {search tool used} — {key finding}
2. {topic} — {search tool used} — {key finding}

## Key Patterns Discovered
- {Pattern name}: {description with code example}

## API Signatures & Options
```{language}
// Exact API signatures from documentation

Gotchas & Edge Cases

• {Gotcha}: {how to avoid it}

Code Examples

// Working examples from documentation

Sources

• {doc title} — {what was learned}

Reliance on Training Data

• {topic}: Relying on training data. No external verification available.

### 4. Training Data Fallback Protocol

If no documentation search tool yields results for a topic:

1. **Document what was searched** — list the queries attempted and tools used
2. **Explicitly state reliance:**
   > "I am relying on my training data for {topic}. External verification was unavailable."
3. **Proceed with caution** — flag the implementation for human review
4. **Prioritize verification** — if the feature is critical, ask the user for documentation links

**Never silently use training data when external verification is available.** The training data cutoff means APIs may have changed, libraries may have deprecated features, and patterns may have evolved.

### 5. Architecture Decision Records

If research reveals a decision involving:
- Choosing between 2+ viable approaches
- Introducing a new dependency or pattern
- Changing existing architecture

Then create an ADR using the **ADR Skill** at `docs/decisions/NNNN-short-title.md`.

## Integration with Agent Workflow

| Agent | How They Use This Skill |
|---|---|
| **Architect** | Research before design decisions, inform ADRs |
| **Backend Engineer** | Research unfamiliar APIs, library patterns |
| **Frontend Engineer** | Research component libraries, CSS patterns |
| **Mobile Engineer** | Research platform APIs, Flutter packages |
| **Database Expert** | Research query patterns, extension capabilities |

## Rule Compliance
- This skill supports the research phase of all builder agents
- Research logs persist in `docs/research_logs/` for cross-session knowledge
- ADRs follow the format defined in the `adr` skill