What we learned from other systems

ASMP is new. The problems are not. Seven analogs inform bootstrap priorities (capped study — not a 50-protocol literature review).

1. BitTorrent

Steal: Portable manifest (.torrent / .asmp.yaml) + swappable tracker (:7700). Announce + lookup.

Skip: DHT federation in v0.1.

Feeds: Reference server extract, capability URIs as magnet links.

Steal: Optional push presence on start; byebye on stop.

Skip: LAN query flooding as primary discovery.

Feeds: Life ring homelab (post-v0.1).

Steal: Resolve by capability (email.ingest), not package name. Best normie analogy.

Skip: Activity lifecycle complexity.

Feeds: Docs positioning, service_find UX.

Steal: Integration registers → entities with domains → automations discover by domain. One registry, many consumers.

Skip: Human-first YAML UI as primary author.

Feeds: Dashboard consumer pattern, registry web UI.

Steal: One bootstrap document — ~/.asmp/host.yaml → registry API → MCP.

Skip: Full OAuth token flows in v0.1.

Feeds: Ambient discovery, global MCP config.

Steal: capabilities.requires as dependencies; validate before register; semver on manifests.

Skip: Immutable package artifacts (services are living processes).

Feeds: JSON Schema, POST validation, mod attach checks.

Steal: Business ring = curated connectors with policy tags and approval gates.

Skip: Hosted middleware as required architecture.

Feeds: Ring 3, Knox hooks, data.sensitivity.

Protocols that won shipped small declaration + lookup tool + client ecosystem:

Build the tool that makes people want the protocol.