IMPORTANT

AI Assist Note (Knowledge Heritage): This document is part of the "Sovereign Reality" documentation.

• @docs ARCHITECTURE:Documentation
• Failure Path: Information drift, legacy terminology, or documentation mismatch.
• Telemetry Link: Cross-reference with execution/parity_guard.py results.

AI Assist Note

Automated governance and architectural tracking.

🔍 Debugging & Observability

Traceability via parity_guard.py.

IMPORTANT

AI Assist Note (Knowledge Heritage): This document is part of the "Sovereign Reality" documentation.

• @docs ARCHITECTURE:Documentation
• Failure Path: Information drift, legacy terminology, or documentation mismatch.
• Telemetry Link: Cross-reference with execution/parity_guard.py results.

🚀 Sovereign Starter Kits

Status: Active
Last Verified: 2026-04-12
Classification: Sovereign

---

Tadpole OS ships with local starter kits for common SMB workflows. Each kit includes a swarm.json definition, agent blueprints, and at least one workflow markdown file.

Location

All built-in kits live in starter_kits/.

Available Kits

1. Marketing & Growth

• Directory: starter_kits/marketing_growth
• Swarm ID: marketing-growth-starter
• Initial Agent: marketing-alpha
• Required Agents: marketing-alpha, content-creator, seo-analyst
• Workflow: workflows/campaign_gen.md
• Goal: Content strategy, campaign generation, and SEO support for SMB growth teams.

2. Customer Success & Triage

• Directory: starter_kits/customer_success
• Swarm ID: customer-success-starter
• Initial Agent: success-lead
• Required Agents: success-lead, triage-bot, feedback-synthesizer
• Workflow: workflows/support_triage.md
• Goal: Ticket routing, feedback analysis, and support-response preparation.

3. Finance & Compliance

• Directory: starter_kits/finance_compliance
• Swarm ID: finance-compliance-starter
• Initial Agent: finance-lead
• Required Agents: finance-lead, expense-auditor, compliance-agent
• Workflow: workflows/fiscal_audit.md
• Goal: Expense review, compliance checks, and financial audit support.

4. Explorer Scout

• Directory: starter_kits/explorer_scout
• Swarm ID: explorer-scout-starter
• Initial Agent: scout-alpha
• Required Agents: scout-alpha, github-analyst
• Workflow: workflows/github_scout.md
• Goal: Repository discovery and competitive codebase scouting.

How To Use

Option A: Install From a Git Repository

The current template-install endpoint expects a repository URL plus a path inside that repository:

POST /v1/engine/templates/install

{
  "repository_url": "https://github.com/DDS-Solutions/Tadpole-OS.git",
  "path": "starter_kits/marketing_growth"
}

What the engine currently does:

• Clones the target repository into data/.bunker_cache/<uuid>/
• Copies agent JSON files into data/swarm_config/agents/
• Copies swarm.json into data/swarm_config/installed/<kit>/
• Copies workflow markdown into directives/
• Merges mcps.json into .agent/mcp_config.json when present
• Persists installed agents into SQLite and the live in-memory registry

Option B: Manual Local Install

1. Copy a starter-kit directory from starter_kits/<kit>/.
2. Place agent JSON files where your runtime expects them, typically data/swarm_config/agents/.
3. Copy any workflow markdown into directives/ if you want it available immediately to the workflow/skill system.
4. Restart the engine or refresh the relevant registries if needed.

Governance Defaults

Starter kits are templates, not trust bypasses. Review these before production use:

• Agent budgets and oversight behavior still depend on your runtime settings and policies.
• Sensitive tool calls should remain approval-gated for SMB deployments.
• Local-first deployments should pair starter kits with PRIVACY_MODE=true, scoped memory, and a real NEURAL_TOKEN.