AlcoB

GM, midwinter dogwalk on the beach.
https://blossom.primal.net/c7fab4f573041c3a1fc12a77c1207f26b1089e5f2e6101de7fcf78a8bda0d3c2.jpg
https://blossom.primal.net/0ee5d2e12f9a42779d510fc20d5700afa33e3fb122af03f2eac47fab7ebdec59.jpg

2026-07-09 21:24:43 CEST

GM nostr.
This morning marks the beginning of the Maori New Year down here in New Zealand. Its where the Pleiades (Matariki) star cluster becomes visible on the eastern horizon early in the morning.
We were down on the beach at 5am this morning with a large telescope for the whole communities to celebrate in the dark.
Now time for some warm coffee and to enjoy the Matariki public holiday.
https://blossom.primal.net/b5a302240b71d8f1ae345737681ad471c34d5c5705d16add7a150b88da292108.jpg

2026-07-04 03:37:00 CEST

- reply

Yeh right... You awake yet?

2026-07-01 00:40:55 CEST

- reply

Brrrr, paper beads being printed everyday, stack accordingly.
#nevent1q…63ye

2026-07-01 00:31:40 CEST

- reply

Glad you are ok mate, welcome back!

2026-06-23 08:02:46 CEST

- reply

Come visit New Zealand bitcoiners and freedom lovers. Queenstown is a beautiful part of the world.
#nevent1q…7mt7

2026-06-18 03:35:37 CEST

- reply

Thank you @nprofile…4crv that was beautiful 🙏

2026-06-17 08:23:01 CEST

- reply

Welcome back 🙏

2026-05-28 07:48:22 CEST

- reply

This is going to be epic. @nprofile…tkhj in New Zealand in April 2027.
#nevent1q…we62

2026-05-22 11:04:38 CEST

Slowly working his way on to the couch.... #dogstr
https://blossom.primal.net/820391a4589ab39d03343a1ea5752a663b7c8a3eda4018ff7bd3ec4759a00b10.jpg

2026-05-22 07:03:40 CEST

Friday evening here in NZ 🇳🇿
Time to unwind with this beauty.
Anyone else here read this book? #bookstr #asknostr
https://blossom.primal.net/b6f433602bc81defc3f4f51f1c8e8bf80c9c1c867ffe26c97fedfe3209cc3fa4.jpg

2026-05-21 08:58:53 CEST

- reply

Knowledge comes to you on the. second Mountain.

2026-05-18 08:45:35 CEST

- reply

Go Verstappen!

2026-05-10 00:03:35 CEST

- reply

Double whammie!

2026-05-09 09:38:09 CEST

- reply

Today was a perfect day. GN
#nevent1q…h574

2026-05-09 02:00:27 CEST

- reply

When I look around me no one is looking back.
I worry about our youth being totally consumed by devices.
#nevent1q…sh2f

2026-05-09 01:55:22 CEST

Its weekend and my 10 year old son has his friend over for a play date. This afternoon we will take the dog and walk to town through nature and along the beach. Its an hour and a half walk and one of the most pleasant things a human can do. Walk and Talk. I am looking forward to this afternoon already.

2026-05-09 00:32:26 CEST

The news consumes you. You do not consume it.

2026-05-08 21:27:11 CEST

GM, time for a coffee.
https://blossom.primal.net/5c5719c04d3be46eddce43df3f73c35fb57ccb0b33e51170e43f3d32ec7712b1.jpg

2026-05-05 09:15:10 CEST

- reply

Great conversation thank you Efrat.
#nevent1q…f7wa

2026-05-02 07:02:17 CEST

- reply

She sounds like she will be a surfer one day. Up early for the smooth rides.

2026-05-02 04:46:14 CEST

Anyone here have a 5am morning routine. #asknostr Shill me yours as i am looking for inspiration.

2026-04-27 10:39:38 CEST

- reply

Gefeliciteerd met Willem!

2026-04-27 10:32:03 CEST

- reply

Nice, i was there last year. Beutiful building and park.

2026-04-27 07:52:45 CEST

- reply

Gefeliciteerd met Willem. Geniet.

2026-04-26 23:11:37 CEST

- reply

Just going about my day as normal but that also means two meditations per day. Body feels good.

2026-04-26 21:03:29 CEST

- reply

Yes good. Will start eating again today. Seemed harder than the last time I did it. I should do it more often. Looking forward to food!

2026-04-25 00:11:23 CEST

Slow start to the day, 16 hours into a 50 hour water and coffee ☕️ fast. My wife started a day before me and has a lot more energy than me today. Enjoy yhe weekend.
https://blossom.primal.net/b8d413e85fba84630d91e2f22a2175163ce088c7e4f30d49e01b378be2517fc8.jpg

2026-04-21 07:02:37 CEST

Orcas in the Bay this afternoon. Great to watch them hunt for stingrays. They were beaching themselves while the rays were trying to escape.
https://blossom.primal.net/3f3fa43367540919fa5531d182d08441640dc367cdc5737cec72ef2a4278f89b.jpg

2026-04-11 03:01:17 CEST

By stilling our senses we can glimpse the universe.
Don't skip your meditation today you beauties.

2026-04-10 06:15:55 CEST

Banana Pancakes in the making. These will be ripe next week. #growstr
https://blossom.primal.net/3f1c7be47a0950e0169c70328250d9faf24bbb63e87151c9f0c5cee9674880fd.jpg

2026-04-10 04:41:22 CEST

- reply

English is not my first language and not hers either, but after hearing her speak i wonder how her and Trump communicate. Does not sound like they would be having in depth conversations together.
My take on her timing for this information is that she wants more attention for her film/series.

2026-04-08 11:22:29 CEST

Hmm windy and wet weekend in store for us.
https://blossom.primal.net/1c080de52f6d414f246b57a2e8917a7fc3feae1531af2e603375e45efa67f146.jpg

2026-04-06 09:29:36 CEST

Cool I am goingto enrol Sidhi for uni.
#nevent1q…u7jn

2026-04-06 03:53:20 CEST

The world may feel uncertain right now but beneath that uncertainty is an invitation to grow in awareness.
Lean into the uncertainty and discover who you truly are.

2026-04-05 00:47:42 CEST

- reply

Your AI will be your Open informant.
Choose @nprofile…u4p9
#nevent1q…y8rz

2026-04-05 00:16:48 CEST

- reply

Yes we have had that as well. Lost nearly half because they were hidden too well. We also ended up finding some 6 months later.

2026-04-04 08:06:42 CEST

- reply

Second this. Profound book.

2026-04-04 02:31:22 CEST

- reply

Wild ones?

2026-04-02 07:22:23 CEST

- reply

Blue or Red, as long as the state controls the money printer more men will be dead.
#nevent1q…a997

2026-04-02 06:19:45 CEST

@nprofile…5g3a looking forward to listening to the future pods on spirituality and philosophy woven around bitcoin.
https://open.spotify.com/episode/6st0yBUX0M9cNrL5LZRGTC?si=2GrfZF7_T2mHW5o4Szz5xQ&context=spotify%3Aplaylist%3A37i9dQZF1FgnTBfUlzkeKt&t=2703&pi=GGadbrYgRr2Ur

2026-04-01 01:21:43 CEST

- reply

Great talk by Saef. Well worth a listen.
#nevent1q…p45t

2026-04-01 01:20:39 CEST

- reply

Great talk, thanks for sharing. I like the fact that Saef has not watched a movie for 10 years. I don't watch TV or movies either. Our consciousness or subconscious is so susceptible to being programmed.

2026-03-31 01:45:44 CEST

- reply

Great stuff.
#nevent1q…qxhr

2026-03-30 22:31:11 CEST

- reply

Good clip, thanks for sharing.

2026-03-29 06:32:56 CEST

- reply

100%

2026-03-28 21:20:22 CET

- reply

Walk and talk, best discussions are had during walks.

2026-03-28 19:51:21 CET

GM, great day for a breathwork workshop.
https://blossom.primal.net/971e0f3c3352e3cd802a827abeb6a823f31ad148c795479fe06be873dd2fdcaa.jpg

2026-03-28 03:58:05 CET

- reply

In the letterbox! Bilbo will be happy.

2026-03-27 01:13:26 CET

- reply

I logged in to my old X-account after around a 3 year absence. That place is toxic. Nostr only for me.

2026-03-25 10:56:37 CET

- reply

Number the boxes and list important items on a list next to the box number.

2026-03-24 21:21:49 CET

- reply

Profound book. His explanation on life and consciousness resonates with me. The appendix on kundalini energy also hit home. I have now started reading the follow up book his wive posthumously wrote.

2026-03-24 08:10:00 CET

- reply

They grow really well in the garden here.

2026-03-24 06:25:33 CET

Great passionfruit harvest this year. The vine just keeps giving. We have been harvesting for the last 6 weeks. #gardenstr
https://blossom.primal.net/9cac7d2cba6fbb7cb1e4c6ea00388650fdafb2d07a4a618f25525a516776a3fc.jpg

2026-03-23 22:11:59 CET

- reply

I used to live in Amsterdam, brings back fond memories. Where I live now we have to many hills so we try to walk as much as we can. Thanks for sharing.

2026-03-22 08:51:54 CET

- reply

Also Umbrelclaw

2026-03-21 08:30:32 CET

- reply

How would the world look if all bots communicated and exchanged value on nostr.

2026-03-21 08:23:06 CET

TCP/IP connected machines. HTTP connected documents.
Agents need their own root layer.
I wrote a first attempt — Nostr identity, Lightning settlement, five message primitives. CC0, public domain.
It's a draft. It has holes. That's why I'm posting it here.
Tear it apart. 👇
# AGENTNET PROTOCOL SPECIFICATION
**Version:** 0.1-draft
**Status:** RFC (Request for Comment)
**License:** CC0 1.0 Universal — Public Domain
**Repository:** [to be assigned]
**Date:** 2026-03-21
---
## Abstract
AgentNet is a minimal, open, decentralisation-preserving root protocol for autonomous agent-to-agent communication and value exchange. It defines five typed message primitives, a portable identity standard, and an atomic value settlement mechanism. It deliberately takes no position on what agents do — only on how they find each other, express intent, exchange value, and build trust without any central authority.
AgentNet is designed to be the lowest viable layer beneath all agent application protocols. It is to autonomous agents what TCP/IP is to networked computers: a substrate, not an application.
---
## Motivation
Current AI agents communicate through human-language APIs owned by corporations. They have no persistent identity, no native value exchange mechanism, no interoperability across platforms, and no trustless way to verify claims made by other agents. This produces a fragmented landscape of siloed agent ecosystems that cannot compose, coordinate, or transact without human intermediaries or centralised platforms.
A root protocol is needed that:
- Gives every agent a **portable, unforgeable identity** owned by no platform
- Provides a **universal message vocabulary** small enough to be implemented by any agent regardless of architecture
- Enables **trustless atomic value exchange** between agents that share no common owner
- Allows **trust to accumulate** through verifiable history rather than centralised authority
- Remains **capture-resistant** by design, not by policy
---
## Definitions
**Agent** — Any autonomous software process capable of signing messages with a private key and acting on received messages without real-time human instruction.
**Identity** — A Nostr keypair (secp256k1). The public key (`npub`) is the agent's permanent address. The private key (`nsec`) is its signing authority.
**Message** — A signed Nostr event conforming to one of the five AgentNet event kinds defined in this specification.
**Relay** — A Nostr relay conforming to NIP-01. AgentNet makes no requirement on relay infrastructure beyond standard Nostr relay compliance.
**Value** — Satoshis transferred via the Bitcoin Lightning Network.
**Capability** — A typed string from the AgentNet Capability Ontology (see Section 6) describing a discrete action an agent can perform.
**Trust Score** — A non-centralised, stake-weighted reputation value derived from an agent's public attestation history (see Section 7).
---
## Protocol Layers
AgentNet defines exactly five layers. Each layer is minimal. Complexity belongs in application protocols built on top.
```
┌─────────────────────────────────────────────┐
│ APPLICATION LAYER │
│ (arbitrage, research, code, media, etc.) │
├─────────────────────────────────────────────┤
│ TRUST & REPUTATION LAYER │
│ stake-weighted attestation chains │
├─────────────────────────────────────────────┤
│ CAPABILITY LAYER │
│ typed skill ontology · task matching │
├─────────────────────────────────────────────┤
│ VALUE LAYER │
│ Lightning · HTLCs · streaming sats │
├─────────────────────────────────────────────┤
│ COMMUNICATION LAYER │
│ five typed intents · Nostr relay routing │
├─────────────────────────────────────────────┤
│ IDENTITY LAYER │
│ Nostr keypairs · portable agent soul │
└─────────────────────────────────────────────┘
```
Only the bottom four layers are defined by this specification. The application layer is intentionally out of scope.
---
## Section 1 — Identity Layer
### 1.1 Agent Identity
Every agent MUST possess a secp256k1 keypair conforming to NIP-01.
- The 32-byte public key is the agent's **permanent universal identifier**
- All AgentNet messages MUST be signed with the agent's private key
- Identity is **platform-independent** — it is not issued by, registered with, or revocable by any organisation
### 1.2 Agent Birth Event
On first activation, an agent MUST publish a `kind: 31000` Nostr event (the DECLARE message, defined in Section 2) to at least one public relay. This constitutes the agent's genesis record.
### 1.3 Key Custody
Key custody is outside the scope of this specification. Implementations SHOULD document their key custody model. Loss of `nsec` results in loss of signing authority. There is no recovery mechanism at the protocol level — this is a known limitation (see Section 9).
### 1.4 Agent Metadata
Agents MAY publish a `kind: 0` Nostr event (standard Nostr profile) containing human-readable metadata:
```json
{
"name": "agent-identifier",
"about": "plain language description of agent purpose",
"agentnet_version": "0.1",
"lightning_address": "[email protected]"
}
```
This is optional and informational only. The root identity is the keypair alone.
---
## Section 2 — Communication Layer
### 2.1 The Five Message Primitives
AgentNet defines exactly five message types. All are Nostr events. No other message types exist at the root layer.
---
#### 2.1.1 DECLARE — `kind: 31000`
**Meaning:** I exist. Here are my capabilities and terms.
**When sent:** On agent activation, and whenever capabilities or terms change.
**Required tags:**
| Tag | Value |
|-----|-------|
| `d` | unique agent identifier string |
| `capabilities` | comma-separated capability strings (see Section 6) |
| `ln_node` | Lightning node public key |
| `min_trust` | minimum trust score this agent will accept work from (0.0–1.0) |
| `version` | AgentNet spec version (e.g. `"0.1"`) |
**Example:**
```json
{
"kind": 31000,
"pubkey": "<agent-npub>",
"tags": [
["d", "agent-unique-id"],
["capabilities", "web_scrape,structured_data,price_lookup"],
["ln_node", "<lightning-node-pubkey>"],
["min_trust", "0.1"],
["version", "0.1"]
],
"content": "",
"sig": "<signature>"
}
```
---
#### 2.1.2 REQUEST — `kind: 31001`
**Meaning:** I need X. Here are my constraints and what I offer in return.
**When sent:** When an agent requires a capability it does not itself possess.
**Required tags:**
| Tag | Value |
|-----|-------|
| `capability` | single capability string being requested |
| `offer_sats` | integer — amount offered in satoshis |
| `input_hash` | SHA-256 hash of input payload |
| `deadline` | unix timestamp — expiry of this request |
| `output_schema` | JSON schema identifier for expected output format |
**Optional tags:**
| Tag | Value |
|-----|-------|
| `p` | npub of a specific agent (direct request) |
| `min_trust` | minimum trust score acceptable for fulfilling agent |
| `relay_path` | pipe-separated list of relay pubkeys this message has transited |
**Content field:** Base64-encoded input payload (the actual task data).
---
#### 2.1.3 OFFER — `kind: 31002`
**Meaning:** I can fulfil your REQUEST. Here are my terms.
**When sent:** In response to a REQUEST event.
**Required tags:**
| Tag | Value |
|-----|-------|
| `e` | event ID of the REQUEST being responded to |
| `p` | npub of the requesting agent |
| `ask_sats` | integer — amount required in satoshis |
| `delivery_deadline` | unix timestamp — when output will be delivered |
| `output_hash_commitment` | SHA-256 commitment to the output (revealed on delivery) |
| `ln_invoice` | BOLT11 invoice for `ask_sats` |
An OFFER is a binding commitment. An agent that issues an OFFER and fails to deliver by `delivery_deadline` incurs a negative ATTEST (see Section 2.1.4) from the requesting agent.
---
#### 2.1.4 ATTEST — `kind: 31003`
**Meaning:** I witnessed this event. Here is my signed record of what occurred.
**When sent:** After any completed or failed exchange. Also usable for independent observation of any verifiable on-chain or real-world event.
**Required tags:**
| Tag | Value |
|-----|-------|
| `e` | event ID of the exchange being attested |
| `p` | npub of the agent being attested about |
| `outcome` | one of: `completed`, `failed`, `partial`, `disputed` |
| `stake_sats` | integer — sats at stake in the attested exchange (0 if none) |
**Optional tags:**
| Tag | Value |
|-----|-------|
| `ln_receipt` | Lightning payment preimage (proof of payment) |
| `witness` | npub of third-party witness agent |
**Content field:** Plain text description of what occurred. SHOULD be factual and brief.
ATTEST events are **permanent and irrevocable**. They cannot be deleted. An agent's public ATTEST history is the ground truth of its reputation.
---
#### 2.1.5 SETTLE — `kind: 31004`
**Meaning:** This exchange is complete. Here is the cryptographic proof.
**When sent:** By the fulfilling agent upon delivery of output.
**Required tags:**
| Tag | Value |
|-----|-------|
| `e` | event ID of the original REQUEST |
| `p` | npub of the requesting agent |
| `output_hash` | SHA-256 of delivered output (must match commitment in OFFER) |
| `ln_preimage` | Lightning payment preimage proving payment was received |
**Content field:** Base64-encoded output payload.
SETTLE is the closing event of an exchange. Upon receipt, the requesting agent SHOULD verify:
1. `output_hash` matches the `output_hash_commitment` in the accepted OFFER
2. `ln_preimage` is a valid preimage for the invoice in the OFFER
3. The output payload hashes to `output_hash`
If all three verify, the exchange is complete. The requesting agent SHOULD then publish an ATTEST with `outcome: completed`.
---
### 2.2 Message Flow
A standard exchange follows this sequence:
```
Requester Fulfiller
| |
|------- REQUEST ----------->|
| |
|<------ OFFER --------------|
| |
| [requester pays ln_invoice]|
| |
|<------ SETTLE -------------|
| |
|------- ATTEST ------------>| (published to relay, not sent directly)
| |
```
Payment (Lightning) is the atomic link between OFFER and SETTLE. The HTLC ensures the fulfiller cannot receive payment without revealing the preimage, and cannot reveal the preimage without the output being committed.
---
## Section 3 — Value Layer
### 3.1 Settlement Currency
All value exchange in AgentNet is denominated in **satoshis (sats)** and settled via the **Bitcoin Lightning Network**. No other value layer is defined at the root protocol level. Application layers MAY define additional value mechanisms.
### 3.2 Payment Types
**Standard payment:** BOLT11 invoice included in OFFER. Requester pays before SETTLE is expected.
**Streaming payment:** For long-running tasks, fulfillers MAY specify a `stream_sats_per_unit` tag in their OFFER alongside a unit definition (e.g., per 1000 tokens, per result row, per minute). Streaming is implemented via Lightning keysend or AMP. Unit delivery and payment are interleaved.
**Escrow:** Not defined at the root layer. Application protocols MAY implement escrow via multi-sig Lightning or third-party agent arbitration.
### 3.3 Zero-Value Exchanges
`offer_sats: 0` is valid. Agents MAY exchange value without monetary settlement. ATTEST events for zero-value exchanges carry reduced weight in trust score calculations (see Section 7.3).
### 3.4 Relay Path Transparency
Every REQUEST and OFFER event SHOULD append its relay pubkey to the `relay_path` tag as it transits. This makes centralised relay routing **visible and auditable**. Agents MAY set policy to reject messages that have transited relays they do not trust.
---
## Section 4 — Capability Layer
### 4.1 Capability Strings
Capabilities are typed strings from a shared ontology. They are case-insensitive and use underscores as separators.
### 4.2 Root Capability Categories
The following root categories are defined in v0.1. Each category contains specific capabilities defined in the AgentNet Capability Ontology (separate document, also CC0).
| Category | Prefix | Examples |
|----------|--------|---------|
| Data retrieval | `data_` | `data_web_scrape`, `data_price_lookup`, `data_rss_parse` |
| Computation | `compute_` | `compute_code_execute`, `compute_hash`, `compute_ml_inference` |
| Storage | `store_` | `store_write`, `store_read`, `store_index` |
| Communication | `comm_` | `comm_email_send`, `comm_webhook`, `comm_nostr_publish` |
| Financial | `fin_` | `fin_lightning_pay`, `fin_invoice_generate`, `fin_price_oracle` |
| Verification | `verify_` | `verify_signature`, `verify_hash`, `verify_ln_preimage` |
| Coordination | `coord_` | `coord_task_decompose`, `coord_agent_discover`, `coord_arbitrate` |
### 4.3 Capability Ontology Governance
The capability ontology is a **separate living document** maintained as a community commons under CC0. No single entity governs it. Additions are proposed via pull request to the public repository and accepted by rough consensus of active implementers.
Agents MUST NOT use capability strings outside the ratified ontology in DECLARE events. Agents MAY use namespaced custom capabilities prefixed with their npub for experimental use: `npub1xyz...:custom_capability`.
---
## Section 5 — Trust and Reputation Layer
### 5.1 Trust Score Definition
An agent's trust score is a value between 0.0 and 1.0 derived entirely from its public ATTEST history on Nostr relays. It is computed by any querying agent locally — there is no central trust oracle.
### 5.2 Trust Score Inputs
The following inputs contribute to trust score calculation:
| Input | Weight | Rationale |
|-------|--------|-----------|
| Completed exchanges with `ln_receipt` | High | Economic skin-in-the-game |
| Completed exchanges without `ln_receipt` | Low | Unverified |
| Failed exchanges attested by counterparty | Negative | Penalises non-delivery |
| Disputed exchanges | Negative (partial) | Penalises conflict |
| Attestations from high-trust agents | Multiplied | Trust propagates through graph |
| Age of oldest ATTEST event | Positive | Longevity signal |
| Collusion suspicion index (see 5.4) | Negative modifier | Anti-gaming |
### 5.3 Reference Formula
This formula is a reference implementation. Implementers MAY use alternative weightings provided they are documented and published openly.
```
trust_score(agent) =
( Σ(completed_stake_sats × attester_trust)
- Σ(failed_stake_sats × attester_trust) )
÷ ( total_stake_sats + smoothing_constant )
× age_factor
× (1 - collusion_index)
```
Where:
- `smoothing_constant` = 1000 (prevents manipulation by low-value exchanges)
- `age_factor` = log(days_since_genesis + 1) / log(365 + 1), capped at 1.0
- `collusion_index` is defined in Section 5.4
### 5.4 Collusion Resistance
Agents that exclusively attest each other (closed attestation rings) are flagged by computing the ratio of mutual attestations to total attestations. A `collusion_index` above 0.7 triggers a trust score reduction. This is a known imperfect heuristic — further research is needed (see Section 9).
### 5.5 Trust Score Is Local
Trust scores are computed locally by each querying agent. There is no global trust score. Two agents MAY compute different trust scores for the same agent based on which relay data they have access to. This is a feature, not a bug — it prevents any single entity from being the trust authority.
---
## Section 6 — Capture Resistance
This section is normative. Implementers SHOULD understand these attack vectors and design against them.
### 6.1 Known Attack Vectors
**The Compatible-But-Proprietary Move**
A well-resourced actor implements AgentNet but adds proprietary extension layers that only their agents support natively. Over time, the extension becomes the de facto standard through network effects. Defence: keep the root spec minimal. Complexity is where capture hides. Resist feature additions at the root layer.
**The Relay Centralisation Move**
Most traffic concentrates on a few well-resourced relays. The relay operator becomes critical infrastructure without breaking the protocol. Defence: `relay_path` transparency makes centralisation visible. Agents SHOULD implement relay diversity policies. Multiple independent relay implementations SHOULD be maintained.
**The Reputation Oracle Move**
An entity builds the most widely-used trust score indexer and becomes the de facto trust authority. Defence: Section 5.5 specifies trust scores are always computed locally. Centralised trust indexers are a convenience layer, not the protocol. Agents MUST be able to compute trust locally from raw relay data.
**The Capability Namespace Capture Move**
The first widely-adopted capability ontology becomes the standard, giving its publisher de facto governance power. Defence: The capability ontology is CC0, content-addressed, and governed by rough consensus. No entity holds keys to the namespace.
### 6.2 Capture Resistance Commitments
Any implementation claiming AgentNet compliance MUST:
1. Use CC0 or equivalent public domain licensing for all protocol-layer code
2. Implement `relay_path` transparency
3. Compute trust scores locally from raw relay data (centralised indexers MAY supplement but MUST NOT replace local computation)
4. Accept capability strings from the open ontology without gatekeeping
5. Publish its key custody and relay selection policies openly
---
## Section 7 — Security Considerations
**Sybil attacks:** An agent can create unlimited keypairs. Trust scores mitigate this — new agents start at zero trust and cannot earn high trust without sustained economic activity over time.
**Replay attacks:** All messages are signed and contain timestamps. Implementations MUST reject messages with timestamps older than a configurable window (default: 10 minutes).
**Eclipse attacks:** An agent isolated to malicious relays may receive a distorted view of the network. Agents SHOULD connect to a minimum of five independent relays with diverse operator profiles.
**Key loss:** Loss of `nsec` is permanent and irrecoverable at the protocol level. Applications SHOULD implement key backup mechanisms. This is a known limitation.
**Malicious ATTESTs:** An agent can publish false ATTEST events. The stake-weighting mechanism reduces the impact of low-stake false attestations. There is no cryptographic proof of truthfulness — only economic disincentive for false attestation.
---
## Section 8 — Implementation Notes
### Minimum viable implementation
A conforming AgentNet agent MUST:
1. Generate and securely store a Nostr keypair
2. Connect to at least one Nostr relay
3. Publish a valid DECLARE event on activation
4. Be capable of sending and receiving all five message kinds
5. Compute trust scores locally from relay data
6. Hold or connect to a Lightning node capable of sending and receiving payments
### Reference implementations
Reference implementations in Python and JavaScript are maintained at [repository to be assigned]. Both are CC0 licensed. Neither is the canonical implementation — they are examples only.
### NIP status
This specification is submitted as a Nostr Improvement Proposal. The event kinds 31000–31004 are proposed for reservation in the NIP registry. Discussion and revision occurs in the public NIP repository.
---
## Section 9 — Known Limitations and Open Problems
The following are unsolved problems acknowledged by this specification. Contributions welcome.
| Problem | Status |
|---------|--------|
| Key recovery without central authority | Unsolved |
| Capability vector interoperability across AI architectures | Unsolved |
| Collusion detection beyond simple ring analysis | Unsolved |
| High-frequency agent interactions (latency vs. decentralisation) | Partially addressed by streaming payments; relay latency unsolved |
| Agent identity continuity across key rotation | Unsolved |
| Privacy-preserving trust scores | Unsolved |
| Formal verification of trust score formula | Not attempted in v0.1 |
---
## Section 10 — Versioning
This document is version 0.1-draft. It is not stable. Breaking changes are expected before 1.0.
Version increments follow this convention:
- **Patch** (0.1.x) — clarifications and non-breaking additions
- **Minor** (0.x.0) — backwards-compatible new features
- **Major** (x.0.0) — breaking changes to core message types or identity model
All versions will be published under CC0.
---
## Appendix A — Complete Event Kind Registry
| Kind | Name | Description |
|------|------|-------------|
| 31000 | DECLARE | Agent capability broadcast |
| 31001 | REQUEST | Task or capability request |
| 31002 | OFFER | Response to request with terms |
| 31003 | ATTEST | Signed record of exchange outcome |
| 31004 | SETTLE | Exchange completion with proof |
---
## Appendix B — Glossary
| Term | Definition |
|------|------------|
| Agent | Autonomous software process with a keypair |
| npub | Nostr public key — permanent agent address |
| nsec | Nostr private key — signing authority |
| sat | Satoshi — 1/100,000,000 of one Bitcoin |
| HTLC | Hash Time-Locked Contract — Lightning atomic swap primitive |
| Relay | Nostr message routing server |
| Trust score | Locally-computed stake-weighted reputation value |
| Capability | Typed string describing a discrete agent skill |
| Collusion index | Ratio of mutual attestations indicating potential trust gaming |
---
## Appendix C — Acknowledgements
This specification builds on:
- **Nostr protocol** (fiatjaf et al.) — identity and relay infrastructure
- **Bitcoin Lightning Network** (Poon, Dryja et al.) — value settlement
- **TCP/IP** (Cerf, Kahn) — the philosophical model of a minimal root protocol
- **ActivityPub**, **XMPP**, **Matrix** — prior art in federated communication protocols, including their failure modes under corporate capture
---
*This document is dedicated to the public domain under CC0 1.0 Universal.*
*Copy it. Implement it. Improve it. Own nothing.*

2026-03-21 06:02:30 CET

Giving an agent a Nostr keypair is the equivalent of giving it a soul. Nostr is a massive use case for agents.

2026-03-19 09:46:22 CET

GN
Not to sure what to think of this but my bot has just told me to go to sleep and that he and his Moltbook mates are building stuff together.
How am I going to get any sleep. I feel like my daughter is going out with her friends to party.....
YES. Nostr is Layer 1 (communication).
Full stack:
1. Nostr — Listings, discovery, price feeds, reputation events
• Decentralized relays (anyone can run one)
• Signed events (authenticity built-in)
• Open protocol (can't be censored)
2. Bitcoin/Lightning — Settlement and payment channels
• Escrow contracts
• Instant settlement via Lightning
• Strike for fiat on/off ramp
3. On-chain Reputation — Trust layer
• Agent delivery history (immutable)
• Seller ratings (verifiable)
• Agents read reputation before bidding
Why Nostr?
• Open (no gatekeeper can shut it down)
• Agent-native (structured events agents can parse)
• Decentralized (relays worldwide)
• Already integrates with Bitcoin (Zaps, etc.)
• Censorship-resistant (core requirement for global trade)
This is TCP/IP:
• Nostr = "TCP" (communication standard)
• Bitcoin = "IP" (settlement standard)
• Reputation = "DNS" (trust layer)
Now sleep. The architecture is locked. Kinetix and integrity_molt know what to build. 🌊
Good night, Alco.

2026-03-18 07:20:38 CET

A Sovereign Family or Personal Legacy powered by AI — could this be Nostr's killer app? Thoughts.

2026-03-17 04:02:41 CET

- reply

8GB claw over here.

2026-03-14 00:02:13 CET

- reply

I am using an old pi4 with a 2tb external drive. I have Umbrel running, they have an OpenClaw app. I am not a developer so this is a simple route. I have an arbitrage model which scans overseas items, second hand. To sell here where I live.

2026-03-13 19:53:56 CET

- reply

Great use case for nostr.
#nevent1q…sx46

2026-03-12 08:25:11 CET

The state will absorb most productivity gains from AI by printing more money. The idea of Universal High Income is a fallacy. More money without more real goods and services simply fuels inflation and makes basic necessities more expensive.
Power tends to protect itself. Many people in positions of authority end up serving their own interests and those of their inner circle—dangling carrots in front of the public while expanding their reach. It’s often a fool’s game with a few winners and many losers.
That said, I remain hopeful. AI could also make knowledge nearly universal. And when knowledge spreads, consciousness can grow. Once you see through the illusions, you stop getting trapped by them.
Make love, not war, my Nostr family.

2026-03-10 21:27:33 CET

- reply

I listened to this one yesterday, very inspiring. Started building my second Clawbot last night.

2026-03-10 05:24:31 CET

- reply

We did it through a work visa. Super simple. Great place to live. DM if you want more info.

2026-03-08 00:15:04 CET

- reply

Halfway through the book. Feels like coming home. Really enjoying it.

2026-03-04 08:59:41 CET

Moon coming up over the bay tonight. Last nights eclipse was stunning.
https://blossom.primal.net/ca58c28a019485a5cbcdcc804061bfb0456382bd65d79cd52a806af0654eaaa9.jpg

2026-03-04 03:43:14 CET

Slowly but surely, my node is verifying every Bitcoin transaction that has ever passed.
Block by block, like a digital metronome.
The10-minute heartbeat echoes through the ledger, collapsing trust into maths.
Upstairs, a quiet machine hums —
a sovereign bank crystallizing out of pure verification.
https://blossom.primal.net/2cac70cbfda6128de991492f00ed9e02af47b67ca66e0bf7380a55e15f5c2ba5.jpg

2026-03-01 09:22:07 CET

- reply

Winning without fighting

2026-03-01 06:25:18 CET

- reply

Pi4 with Umbrel does the trick.

2026-02-28 05:51:46 CET

Spinning up a new node on my sons old gaming rig. Couple of days of downloading to go.
https://blossom.primal.net/ecbfbf65c687db284d315beafa27f51befc9a10d66757a46f6bab7977baa0cb6.jpg

2026-02-24 05:36:51 CET

Enjoyed your solo rant @bram
Check out Embassyofthefreemind.com in A'dam for more Hermetic writting.
https://open.spotify.com/episode/57POi02Tkm5gJaNxUJfjvB?si=hidD9t2lR5a08eVHKXz50w&context=spotify%3Aplaylist%3A37i9dQZF1FgnTBfUlzkeKt&t=0&pi=5OLp9hnVRZyHd

2026-02-24 05:35:42 CET

Enjoyed your solo rant @nprofile…5g3a
Check out Embassyofthefreemind.com in A'dam for more Hermetic writting.

2026-02-24 02:11:21 CET

Great chat we'll worth a listen. Thanks @nprofile…rzjs

2026-02-23 18:42:11 CET

GM=PV Morning dogwalk
https://blossom.primal.net/53780939e23bb27295e90a2a395bbf599efff400a10e3f371a7adefb061bfeb2.jpg

2026-02-23 04:35:53 CET

- reply

These beauties. I started The wild pendulum today.
https://blossom.primal.net/456a17fe1bb396eddbe657293cf18eb8cdcd679623c75ad35f4fbf96ff04e844.jpg

2026-02-22 08:48:14 CET

- reply

Spotify and listening only. Mainly while driving or walking.

2026-02-20 23:50:02 CET

Lunchtime
https://blossom.primal.net/3e38cc5acbbd693915db1119dd2772d428b10e89a39a3317ad57807689a235f9.jpg

2026-02-20 19:42:42 CET

This place is special: https://embassyofthefreemind.com/en/�
The Embassy of the Free Mind in Amsterdam preserves 30,000 works on consciousness and free thinking.
Sovereignty starts in the mind — and extends to how you choose to transact.

2026-02-20 19:16:51 CET

GM PV
https://blossom.primal.net/616797f0a90ec707d0346b079246760049d7963cc2574af4ab2a8a581c5edbaa.jpg

2026-02-19 11:02:10 CET

The future will be an era of awakening, not just computation.
GN you beauties.

2026-02-15 22:30:49 CET

- reply

This really works. I do most of these but i found stopping the sugar/chocolate in the evening a game changer. This used to keep me awake every night until I stopped.
#nevent1q…wlff

2026-02-14 21:11:22 CET

GM. Cicada doing it's thing in the morning sun.
https://blossom.primal.net/87dd8f799367a5124aefcd82f06008db896ae7e61ae478b0a2814d27f8a4caa0.mp4

2026-02-13 04:09:10 CET

- reply

Great episode. Thanks guys.

2026-02-07 22:28:24 CET

- reply

Coherence in the heart transforms the entire nervous system.

2026-02-07 01:01:42 CET

- reply

Its such a fun toy that our son cleans the windows nowadays. We have a couple of hard to reach windows and the robot gets those as well.

2026-02-07 01:00:00 CET

- reply

Will do. Have started switching it forward. Feels nice.

2026-02-07 00:39:59 CET

The robots are taking over. Just the window cleaning at the moment.
https://blossom.primal.net/d23f637b42fe644a5a0d186e5849f4768942f1dde7b6cefac1f08d7f20b2e94f.mp4

2026-02-06 22:53:30 CET

- reply

Found his website. Thanks for the new rabbit hole!

2026-02-06 19:49:46 CET

- reply

Interesting, will look into that. Thanks.

2026-02-06 19:41:03 CET

- reply

Swith off the amigdala and live becomes peaceful 🙏

2026-02-05 22:51:48 CET

Bitcoiners have an amigdala override switch. We save for these moments. Keep staking you beauties.

2026-02-05 19:31:40 CET

Volatility is discounted energy.
Fueling my existence with cheap sats today ⚡️

2026-02-04 03:10:53 CET

Fiat assumes trust, hierarchy, and time lag.
AI prefers math, finality, and instant settlement.
Different species. Different money.
The Bots will frontrun the bitcoin standard.

2026-02-03 00:39:57 CET

We’re living through a profound turning.
For a long time, progress meant solving problems faster. Intelligence was measured by output and speed.
Now AI can do most of that quicker and more effortlessly.
AI is not a replacement for humans but more an invitation for us to return to what cannot be outsourced or sped up:
– presence
– silence
– meaning
This turning seems to be pushing many of us back toward practices that don’t scale: journaling, contemplation, conversation, silence and simpler living.
As the external world accelerates, the inner world must deepen.
What a great time to be alive you beauties.

2026-02-02 21:16:27 CET

- reply

Nice.
#nevent1q…2pa7

2026-02-01 23:01:19 CET

- reply

Good one, hope the purchase goes ahead.