SDD Implement Command

When invoked, follow these steps:

Step 0: Flag Parsing and Session Check

Load TOML Defaults

Call the environment tool to read configuration (returns both implement and git sections):

mcp__plugin_foundry_foundry-mcp__environment action="get-config"

This returns:

{
  "success": true,
  "data": {
    "sections": {
      "implement": {"auto": false, "delegate": false, "parallel": false, "model": "haiku"},
      "git": {"enabled": true, "commit_cadence": "task", ...}
    }
  }
}

Use implement values as the baseline. CLI flags override TOML values. The git section is available for commit cadence decisions during implementation.

If the config file is missing or section not found: Use defaults (all false for implement).

Parse Command Flags

Four flags that can be combined:

Flag	Effect
--auto	Skip prompts between tasks (autonomous execution)
--delegate	Use subagent(s) for implementation
--parallel	Run subagents concurrently (implies --delegate)
--model <haiku|sonnet|opus>	Model for delegated tasks (default: haiku)

Resolution order: TOML defaults → CLI flags (CLI wins)

Resulting modes from combinations:

Flags	Subagent	Concurrent	Prompts	Description
(none)	❌	❌	✅	Interactive, inline
--auto	❌	❌	❌	Autonomous, inline
--delegate	✅	❌	✅	Interactive, sequential subagent
--auto --delegate	✅	❌	❌	Autonomous, sequential subagent
--delegate --parallel	✅	✅	✅	Interactive, concurrent subagents
--auto --delegate --parallel	✅	✅	❌	Autonomous, concurrent subagents

Note: --parallel without --delegate implies --delegate.

Check for Existing Session

Before proceeding, check if a paused session exists:

mcp__plugin_foundry_foundry-mcp__task action="session-config" spec_id={spec-id} command="status"

If session is paused:

AskUserQuestion:
"Found paused session (reason: {pause_reason}). Last task: {current_task}"
Options:
- "Resume session" → session-config command="resume", then continue
- "Start fresh" → session-config command="end", then proceed normally
- "Exit"

If session is active: Warn user that a session is already running. Offer to view status or exit.

If session is idle or no session: Proceed to Step 1.

Interactive Fallback (No Flags and No TOML Defaults)

When no CLI flags provided AND TOML defaults are all false, offer mode selection:

AskUserQuestion:
"Select execution mode:"
Options:
- "Interactive, inline (default)" → Proceed to Step 1
- "Autonomous, inline (--auto)" → Proceed with --auto
- "Interactive, delegated (--delegate)" → Proceed with --delegate
- "Autonomous, delegated (--auto --delegate)" → Proceed with --auto --delegate
- "Autonomous, parallel (--auto --delegate --parallel)" → Proceed with all flags

If TOML has defaults set: Skip this prompt and use TOML values directly.

Step 1: Check for Active Specifications

Use mcp__plugin_foundry_foundry-mcp__spec action="list" with status "active" to find specifications in progress.

Step 2: Route Based on Results

If active specs exist:

Pass the active spec context to the sdd-implement skill. The skill will proceed directly to task selection - it does NOT need to re-check for active specs.

IMPORTANT: Do NOT invoke this command or skill recursively. The skill handles the full workflow.

Skill(sdd-implement) "Active spec detected via /implement command. Skip Step 3.1 (spec detection) and proceed directly to Step 3.2 (task selection)."

If no active specs exist:

Check for pending specs using mcp__plugin_foundry_foundry-mcp__spec action="list" with status "pending".

If pending specs found: Ask user if they want to activate one using AskUserQuestion.

If no specs at all: Inform user no specifications exist and offer options:

1. Create a new spec using the sdd-plan skill
2. View completed/archived specs
3. Exit

Step 3: Hand Off to Skill

Once routing is determined, the sdd-implement skill handles the full workflow:

• Task preparation and context gathering
• Implementation guidance
• Progress tracking and verification

This command is the user entry point; the skill contains the detailed workflow logic.