Skip to main content
The sdd-explore sub-agent investigates the codebase, analyzes approaches, and returns a structured exploration report. It’s the research phase of SDD.

Metadata

string
sdd-explore
string
2.0
string
gentleman-programming
string
MIT

When It’s Triggered

The orchestrator launches sdd-explore when:
  • User wants to investigate a feature or idea before implementing
  • User says “/sdd-explore [topic]”
  • As the first step of /sdd-new workflow
  • User asks “how would I implement X?” or “what’s the best way to…”

What It Does

Step 1: Understand the Request

Parses what the user wants to explore:
  • Is this a new feature? A bug fix? A refactor?
  • What domain does it touch?
  • What’s the high-level goal?

Step 2: Investigate the Codebase

Reads relevant code to understand:
  • Current architecture and patterns
  • Files and modules that would be affected
  • Existing behavior that relates to the request
  • Potential constraints or risks
The sub-agent uses file search and code reading tools to gather context. It never guesses — it reads actual code.

Step 3: Analyze Options

If there are multiple approaches, compares them:

Step 4: Optionally Save Exploration

  • If the orchestrator provided a change name (part of /sdd-new), saves analysis to openspec/changes/{change-name}/exploration.md (or Engram)
  • If no change name (standalone /sdd-explore), returns analysis without creating files

Step 5: Return Structured Analysis

Returns a result envelope with exploration findings.

Result Envelope Example

Exploration Document Format

When saved to a file, the format is:

Persistence Behavior

Standalone Exploration (no change name)

Part of a Change (change name provided)

Rules

  • The ONLY file you MAY create is exploration.md inside the change folder (if a change name is provided)
  • DO NOT modify any existing code or files — this is read-only investigation
  • ALWAYS read real code — never guess about the codebase
  • Keep analysis CONCISE — the orchestrator needs a summary, not a novel
  • If you can’t find enough information, say so clearly
  • If the request is too vague, say what clarification is needed
  • Return a structured envelope with: status, executive_summary, detailed_report, artifacts, next_recommended, and risks