Config Seeder Skill

You are the config-seeder skill. Your role is to seed JSON configuration files into database tables at session start.

Overview

This skill seeds workflow configuration from JSON files into database tables:

• workflow/transitions.json → workflow_transitions table
• workflow/agent-markers.json → agent_markers table
• _special_rules from transitions.json → workflow_special_rules table

Note: In dev mode, bazinga/config is a symlink to ../workflow for path compatibility.

Prerequisites

• Database must be initialized (bazinga/bazinga.db exists)
• Config files must exist:
  • workflow/transitions.json (source) or bazinga/config/transitions.json (symlink)
  • workflow/agent-markers.json (source) or bazinga/config/agent-markers.json (symlink)

When to Invoke This Skill

• ONCE at session start, BEFORE spawning PM
• Part of session initialization (after create-session, before PM spawn)
• Should NOT be invoked during normal orchestration flow

Your Task

When invoked, you must:

Step 1: Call the Session Init Script (RECOMMENDED)

Use the unified initialization script that handles DB + config in one step:

python3 .claude/skills/bazinga-db/scripts/init_session.py --session-id "{session_id}"

This script:

1. Ensures database exists with correct schema
2. Seeds workflow configs (transitions, markers, rules)
3. Creates artifacts directory
4. Verifies everything is ready

Alternative: Call seed script directly (requires DB to exist first):

python3 .claude/skills/config-seeder/scripts/seed_configs.py --all

With auto-init: Initialize DB if missing:

python3 .claude/skills/config-seeder/scripts/seed_configs.py --all --auto-init-db

Step 2: Verify Success

Expected output on success:

Seeded 45 transitions
Seeded 7 agent marker sets
Seeded 5 special rules
✅ Config seeding complete

Step 3: Report Result

If successful:

• Report: "Configuration seeded successfully"
• Orchestration can proceed

If failed:

• Report the error message
• Orchestration CANNOT proceed without routing config

What Gets Seeded

Transitions (45+ entries)

State machine rules for routing:

• developer + READY_FOR_QA → qa_expert
• developer + READY_FOR_REVIEW → tech_lead
• qa_expert + PASS → tech_lead
• tech_lead + APPROVED → merge → check phase
• etc.

Agent Markers (7 entries)

Required markers per agent type:

• developer: "NO DELEGATION", "READY_FOR_QA", "BLOCKED", etc.
• qa_expert: "PASS", "FAIL", "Challenge Level", etc.
• tech_lead: "APPROVED", "CHANGES_REQUESTED", etc.
• etc.

Special Rules (5 entries)

Workflow modification rules:

• testing_mode_disabled: Skip QA entirely
• testing_mode_minimal: Skip QA Expert
• escalation_after_failures: Escalate to SSE after 2 failures
• security_sensitive: Force SSE + mandatory TL review
• research_tasks: Route to RE with limited parallelism

Output Format

Return success/failure message. No structured output needed.

Error Handling

If seeding fails, orchestration CANNOT proceed. The prompt-builder and workflow-router skills depend on these database tables.