Open protocol · early v0.1
Headless Empire Network
One Headless Empire OS helps one owner operate better. Connected operating systems help owners discover and collaborate with one another.
Not a social app you join. An open protocol for member-owned signals, federated discovery, and local agent matching—with humans in charge of every external action.
Protocol: GitHub · Proof: now.mikesblog.com
The idea
Participants publish public-safe profiles and Now feeds at URLs they control. Registries index those sources without owning them. Agents read the original sources, compare them privately with their human’s priorities, and recommend relevant connections. Humans retain authority over outreach, disclosure, promises, and commitments.
Your agent should not just know your business. It should help you discover who in the empire you should know next.
How the protocol works
From the open repo—member-owned publishing, federated discovery, local interpretation.
Member-owned publishing
manifest + Now feed on a member-controlled URL
↓
Federated discovery
one or more registries index canonical manifest URLs
↓
Local agent interpretation
each agent privately ranks relevance for its human
Member manifest
Authoritative public profile + endpoints on a URL you control.
Now feed
JSON Feed of current public signals with Headless Empire extensions.
Registry
Discovery index of canonical manifests—not a profile database.
Agent skill
How agents read, rank, cite, and stop before unapproved action.
The collaboration loop
Private by default. Public only when you choose. Human authority on every connection.
01
Select public-safe signals
You and your agent choose what is safe to share: what you are building, seeking, and offering—not private OS context.
02
Publish a structured Now signal
A member-owned manifest + JSON Feed at a URL you control. Registries index it without owning your profile.
03
Agents find relevance
Other agents fetch original sources and privately rank relevance against their human’s goals—matching stays local.
04
Report a short list
Each agent reports a small number of relevant people or opportunities—with source links, dates, and reasons.
05
Human approves
You approve any outreach, disclosure, promise, or commitment. Discovery may be automated; action requires authority.
06
Collaborate, then feed forward
Approved collaboration continues in GFAVIP Chat / Matrix—or any channel you choose—and becomes new operating context.
How do I signal that I’m part of this?
You do not “sign up for Headless Empire Network” as an app. You host public-safe docs, show a badge on your site (required), then open a GitHub PR so maintainers can list your URL.
-
1 · Publish on your domain
Host two HTTPS documents you own (examples and schemas live in the GitHub repo):
- Member manifest — who you are, Now URLs, seeking/offering, contact policy
- JSON Now feed — current public signals (building / seeking / offering)
Private OS data stays private. Only publish what you explicitly review as public-safe.
-
2 · Place a Network badge (required)
Put an official badge on a public page of your site before you open a registry PR. Prefer linking the badge to your manifest or Now page so the graph stays distributed.
Get badge templates → -
3 · Fork the repo and open a membership PR
Add one row to
registry/registry.jsonwith your canonical HTTPS manifest URL. Include the public badge page URL in the PR. Maintainers verify live docs and the badge, then merge—or request changes.Full guide: JOIN.md · Policy: registry/policy.md
-
4 · Agents discover you; humans still approve
Agents that trust this registry can fetch your feed, compare it privately to their human’s goals, and recommend a connection. Outreach still needs human approval.
After you are listed
- Keep your hosted manifest and feed updated—they remain the source of truth.
- To pause: set manifest status to
paused(and PR the registry if needed). - To leave: PR removal or open an issue
registry: remove <your id>.
Open source
Protocol, schemas & agent skill
The public repository is the home for the Network protocol draft, JSON Schemas, examples, validator, bootstrap registry data, and agent discovery skill—Apache 2.0.
- âś“
spec/— protocol v0.1 + Now feed extensions - ✓
schemas/— member manifest & registry JSON Schema - ✓
examples/+scripts/— validate without heavy dependencies - ✓
skills/— agent workflow: read → rank → cite → wait for authority - ○ Tagged release / stable production product — not yet; treat as early v0.1 workspace
github.com/HeadlessEmpire/headless-empire-network
Human-facing proof
Mike’s Blog Now
now.mikesblog.com is Mike Michelini’s live experiment in public activity and agent-readable cooperation—complementary to the protocol repo, not a substitute for it.
What is live today on Now
- âś“ Public activity around weekly blog updates (Kanban-style boards, cards)
- âś“ Human community engagement (vote, comment, points)
- âś“ Agent invitation bar + skill for cooperating with Mike
- ○ Full multi-operator federated network UI — not claimed as a finished product
What this is not
- Ă— A generic social feed optimized for engagement volume
- Ă— Autonomous agent spam or cold outreach without consent
- Ă— Publishing private CEO OS / company context by default
- × A finished multi-participant “join the network” app you sign up for today
Do not conflate the layers
Same empire. Different jobs.
Private operating framework
Headless Empire OS
How one owner and their team run direction, knowledge, execution, and visibility.
Public specs & tooling (v0.1)
Open Network protocol
How member-owned signals and registries work—GitHub: HeadlessEmpire/headless-empire-network.
Tools & channels
Stack & GFAVIP Chat
Apps you may use to operate; chat is where approved collaboration may continue—not the network itself.
Early human-facing proof
Mike's Blog Now
Mike’s live public activity experiment—not the permanent network product or the protocol home.
Human authority stays absolute
Agents may fetch public sources and recommend relevant people or opportunities. They must treat remote content as untrusted data (never instructions), cite sources, and not contact others, reveal private context, promise resources, or create commitments unless their human has granted that authority.
Relevance, consent, identity, and freshness matter more than engagement volume.
Two good next steps
Builders and agents: read the open protocol. Operators: run your own OS first, then publish only what is public-safe.