---
name: alpha-agent-platform
description: Operate and extend the ALPHA Autonomous Agent Intelligence Platform — an agentic AI mesh, asynchronous DAG workflow editor, A2A messaging hub, swarm manager, MCP registry, SSD memory store, LLMOps console, self-healing console, and personal assistant hub. Use when designing agents, workflows, swarms, or UI modules for ALPHA.
version: 0.1.0
license: MIT
author: Adonis Jimenez (@donny-devops)
---

# Skill: ALPHA Agent Platform

## When to use

- Adding or modifying one of the ten ALPHA modules
- Designing an agent, swarm, DAG workflow, or MCP integration to be represented in ALPHA
- Writing documentation or policy for agent behavior, memory, security, or commerce

## Capabilities

| Capability | Module | Entry point |
|-----------|--------|-------------|
| Fleet overview | Command Center | `components/pages/dashboard-page.tsx` |
| Mesh topology & node inspection | Agent Mesh | `components/pages/agent-mesh-page.tsx` |
| Workflow design | DAG Editor | `components/pages/dag-editor-page.tsx` |
| Agent-to-agent messaging | A2A Comms | `components/pages/a2a-comms-page.tsx` |
| Collective orchestration | Swarm Manager | `components/pages/swarm-manager-page.tsx` |
| Tools, APIs, databases | MCP Registry | `components/pages/mcp-registry-page.tsx` |
| Layered memory | Memory Store | `components/pages/memory-store-page.tsx` |
| Models, tokens, cost, evals | AgenticOps | `components/pages/agentic-ops-page.tsx` |
| Detection & remediation | Self-Healing | `components/pages/self-healing-page.tsx` |
| NL control & agent builder | Assistant Hub | `components/pages/assistant-hub-page.tsx` |

## Procedure: add a new module

1. Add the section id to `NavSection` in `nextjs_space/lib/types.ts`.
2. Add the sidebar entry in `components/alpha-sidebar.tsx`.
3. Create `components/pages/<name>-page.tsx` wrapped in `PageWrapper`.
4. Register it in `components/alpha-shell.tsx` via `next/dynamic` with `ssr: false`.
5. Add mock data to `lib/mock-data.ts` with explicit types.
6. Update README module table, ARCHITECTURE.md, and CHANGELOG.md.

## Procedure: define an agent

Follow the agent card format in [AGENTS.md](AGENTS.md): id, role, capabilities, model, tools (allow-list), memory
namespaces, budgets, escalation policy.

## Guardrails

- Every loop must declare `maxIterations`, `tokenBudget`, and `timeoutMs`.
- Tools are deny-by-default; list them explicitly per agent.
- Irreversible actions (payments, production changes, outbound messages to humans) require human approval.
- Do not label a security control as active in UI or docs unless it is implemented.

## References

[ARCHITECTURE.md](ARCHITECTURE.md) · [AGENTS.md](AGENTS.md) · [MEMORY.md](MEMORY.md) · [SECURITY.md](SECURITY.md) ·
[docs/AGENTIC_COMMERCE.md](docs/AGENTIC_COMMERCE.md)