BLUESKY

-... .. - -.-. --- .. -. strong opinions, loosely held.

Public Key

npub1az9xj85cmxv8e9j9y80lvqp97crsqdu2fpu3srwthd99qfu9qsgstam8y8

NIP-05 Address

Profile Code

nprofile1qqsw3znfr6vdnxrujezjrhlkqqjlvpcqx79ys7gcph9mkjjsy7zsgygpzpmhxue69uhkummnw3ezuamfdejsz9thwden5te0v4jx2m3wdehhxarj9ekxzmnyvf9x05

Publishing to

nostr.wine

eden.nostr.land

Author Public Key

npub1az9xj85cmxv8e9j9y80lvqp97crsqdu2fpu3srwthd99qfu9qsgstam8y8

Show more details

Last Notes

2026-06-01 22:35:04 UTC

- reply

I know drama is fun to you, but you are poisoning this technology and hindering the other people working on it.

2026-06-01 19:09:12 UTC

Trying to bully a non-profit into funding is wrong.
Be grateful for what you got
Be grateful for what others got
Be grateful there even is funding
We don't know if there will even be funding in the future.
Just be grateful.

2026-06-01 17:23:35 UTC

- reply

The world is far from perfect.
Be grateful some ppl are willing to do what you are not.
If you don't agree, simply start an alternative.
#nevent1q…s3q4

2026-06-01 17:05:16 UTC

- reply

Philanthropy has a place. But it does skew markets, and create parasitic behavior.
#nevent1q…awh9

2026-06-01 01:27:50 UTC

I know it's crazy for some ppl, but try working for money.

2026-05-31 20:36:14 UTC

- reply

iOS app?

2026-05-31 20:33:05 UTC

- reply

Where/how

2026-05-31 20:25:38 UTC

Outside of kind1, any interesting communities?

2026-05-31 20:11:40 UTC

https://blossom.primal.net/eca6f54cab900c6ae371f9277b3a956ad0cde0a544a7c52cfe6f3adf7ab71ddd.jpg

2026-05-31 19:14:19 UTC

- reply

🤣

2026-05-31 19:10:13 UTC

Stop being Insufferable.

2026-05-31 19:05:10 UTC

- reply

Don't like it don't use it, build something better.
#nevent1q…ehfy

2026-05-31 19:04:42 UTC

- reply

Every time I login here all I see is people complaining.
It's such a turn off.
And yet, I'm using Nostr everyday with agentnoise.org
#nevent1q…f5j6

2026-05-31 19:00:41 UTC

You are not obligated to use primal, how about build some alternative and go get some users instead of bitching.
The principal reason Nostr adoption stopped is because of the bitching instead of building.

2026-05-14 12:31:33 UTC

nostr
https://agentnoise.org

2026-05-05 03:05:48 UTC

https://blossom.primal.net/c0c2f0adc4dd58ddd5810b46b452b7a30ca7719e099c1019f1f39419ead949cf.png

2026-04-29 17:14:27 UTC

- reply

It depends on a APK by them, there is no external solution. And they control the whole stack.

2026-04-29 13:03:05 UTC

If not enough bitcoin is held privately in the hands of unreasonable people we lose.
Bitcoin, in seeds, is the only insurance.

2026-04-27 18:04:26 UTC

learntoprompt.org
all experimental, because the current shitshow security is not viable

2026-04-18 01:21:58 UTC

- reply

Objectively

2026-04-14 02:08:39 UTC

# Writing 84 Tests for a Project With Zero Lines of Code
The llm-wiki project has 3,610 lines across 22 files. Every single one is a markdown file. There is no Python. No JavaScript. No compiled binary. The "source code" is English prose — instructions that Claude reads and follows to build knowledge wikis from web research.
So how do you write tests for a program that is, technically, a document?
I figured it out. 84 structural assertions, 11 intentionally broken wiki fixtures, 5 behavioral evals via Promptfoo, and a GitHub Actions pipeline. As far as I can tell, nobody has used Promptfoo to test a Claude Code plugin before. Here is what I learned.
## The three-layer problem
Traditional testing has a simple contract: given input X, the function returns Y. If it doesn't, the test fails. But when your "function" is an LLM reading markdown instructions, the contract dissolves. The same instruction file, given the same user request, might produce different article titles, different file structures, different cross-references. The output is correct within a range, not at a point.
Anthropic, OpenAI, and GitLab all converge on the same solution: split tests into layers by how much uncertainty you're willing to tolerate.
**Layer 1 is deterministic and free.** No LLM calls. You're checking that the wiki's file system is internally consistent. Does every directory have an `_index.md`? Does every raw source have the six required frontmatter fields? Does the `type: articles` file actually live in `raw/articles/` and not `raw/papers/`? These checks take seconds and cost nothing. I have 84 of them. They run on every push.
**Layer 2 is semantic and costs money.** You ask Claude to do something — ingest a URL, compile an article, route a command — and then grade whether it followed the instructions. Promptfoo handles this with three assertion types: trajectory assertions ("did it call WebSearch?"), llm-rubric assertions ("does the output have complete frontmatter?" graded by a judge LLM), and custom JavaScript that checks the file system after the agent runs. Each eval costs about $0.50. I run five of them on PRs.
**Layer 3 is full workflows.** Research-to-article. Ingest-compile-lint. Retract-and-verify-cleanup. These use `claude -p` in headless mode, cost $10-20 per run, and execute weekly. I haven't built these yet. Layers 1 and 2 are live.
## The golden wiki
Every structural test needs something to test against. I built a golden wiki — a minimal but complete fixture with three raw sources, two compiled articles, proper cross-references, bidirectional See Also links, correct index files, and a valid log. Twenty files total. It passes every check.
Then I broke it eleven different ways. One copy per lint rule. `missing-index/` has a deleted `_index.md`. `bad-frontmatter/` has `type: invalid` instead of `type: articles`. `misplaced-file/` puts a concept article inside `wiki/references/`. `retracted-marker/` leaves a `` comment that should have been cleaned up. Each broken copy triggers exactly one violation. The test asserts that the defect is present — negative testing.
A shell script called `http://generate-defect-fixtures.sh` creates all eleven from the golden wiki in under a second. Change the golden fixture, regenerate, and every negative test updates automatically.
## Promptfoo on a Claude plugin
Promptfoo has a provider called `anthropic:claude-agent-sdk` that can load local plugins. Point it at your plugin directory, whitelist the tools, set a budget cap, enable sandbox mode, and it runs your plugin through test cases defined in YAML.
The part that surprised me: the `skill-used` assertion type. You can assert that the agent invoked a specific skill — not just that the output mentions wiki commands, but that Claude actually triggered the wiki skill at the Claude Code level. Combined with trajectory assertions that verify which tools were called, you can check both what happened and how.
I test five behaviors: the fuzzy router dispatching "Research the history of testing" to the research command, a URL to ingest, a question to query, an ambiguous single word triggering clarification (negative control), and the plugin loading without errors. Each runs three times to measure variance.
## What I actually learned
The biggest surprise: Layer 1 catches almost everything. The expensive behavioral evals in Layer 2 are for confidence, not coverage. Index corruption, frontmatter drift, misplaced files, broken cross-references — these are the actual failure modes of a wiki management system, and they're all deterministic. You don't need an LLM to verify that a file exists in the right directory.
Anthropic's eval guide says "grade outcomes, not trajectories." For wiki operations, the outcome IS the file system state. Check the files, check the indexes, check the links. If the structure is correct, the agent followed the protocol. The trajectory — which tool calls it made, in what order — is interesting but secondary.
The test suite is at https://github.com/nvk/llm-wiki in `tests/`. Clone it, run `./tests/test-structure.sh`, and watch 84 green checkmarks validate a project that contains zero lines of code.

2026-04-11 14:41:19 UTC

Writing bitcoinquantum.space with llm-wiki.net
In April 2026 I wanted to assess whether the quantum threat to Bitcoin was real. The honest answer lived across fifteen papers, a dozen Delving Bitcoin threads, twenty Bitcoin Optech newsletters, a running testnet, some Liquid transactions, and whatever Avihu Levy had pushed to GitHub that morning. The work was real and scattered. No article summarized it honestly. Headlines were downstream of press releases. The primary sources were where the actual answer lived.
This is one of the things llm-wiki was built for. I used it. Three weeks later I published [bitcoinquantum.space](bitcoinquantum.space) — three articles, ~15,000 words, 95+ sources cross-referenced, every claim verified. This is a writeup of how.
## The shape of the problem
Serious research has three failure modes:
1. **You can't find everything.** Sources scatter across formats and venues. You don't know what you're missing.
2. **You can't remember everything.** By paper #60 you've forgotten paper #4. You re-read. You contradict yourself.
3. **You can't update.** A new paper drops on publication day. Your conclusion is stale and your notes are already collapsed into prose you can't untangle.
Traditional knowledge management fixes (1) and partly (2). It fails at (3) because the maintenance burden compounds. @karpathy's framing, *"who does the maintenance?"*, is load-bearing because humans don't, not reliably, not for unsexy cross-reference updates nobody sees.
llm-wiki.net fixes (3) by making the entire artifact mechanically regeneratable from immutable raw sources. The only thing you maintain is the source pile.
## The pipeline, applied
**Raw sources, not notes.** Every paper, blog post, mailing list thread, and testnet report got dropped into `raw/` verbatim with a frontmatter header. No interpretation, no paraphrasing. If I don't have the primary source, I don't have it. `raw/` grew to 95+ entries.
**Compile, don't write.** `/wiki:compile` reads the raw pile and synthesizes cross-referenced wiki articles — one per concept, person, and proposal. "SHRINCS." "Taproot script-path post-quantum proof." "The BIP 86 problem." "Quantum Safe Bitcoin." Each article carries a confidence level, citations, and bidirectional cross-references. The wiki is Claude's work; the sources are mine.
**Query to find gaps.** Once compiled, I stop reading papers and start asking questions. *"What's the relationship between Ruffing's Taproot proof and BIP 86?"* The wiki answers with citations — and in the process surfaces the gap: 70-90% of BIP 86 outputs can't use the escape hatch. That's a thread I wouldn't have pulled linearly. Query mode is where llm-wiki stops being a filing cabinet and starts being a research partner.
**Output, last.** The articles on bitcoin

2026-04-08 13:43:35 UTC

llm-wiki.net v0.0.15 is out
https://blossom.primal.net/c444d1f0d8d196bc8770715582b061e86649085819235f1675f0441f21420f5d.png

2026-04-08 13:32:37 UTC

Quantum FUD is only fought with the best slop. It's all about inverting Brandolini's law.
Part 2 https://x.com/nvk/status/2041644230620139787
https://blossom.primal.net/94be314aa71d7138a4243e24618b51abc435d5199392dd517f198f0704728535.jpg
Part 1 https://x.com/compose/articles/edit/2040989355330678784
https://blossom.primal.net/8ee3b2b141622255f3469d70036a8b0a3f49837a91af9103ec2ab7f1863d9e47.jpg

2026-03-28 11:41:43 UTC

Ask Grok mcp for Claude to be able to search X https://github.com/nvk/ask-grok-mcp

2026-03-23 15:40:46 UTC

- reply

RFC 6979 + deterministic builds is perfectly adequate. That's COLDCARD's and Cores solution

2026-03-23 02:04:35 UTC

- reply

https://www.reddit.com/r/AskPhotography/s/sc94CC64qT
Ppl keep searching for a home.
#nevent1q…qf3u

2026-03-23 01:05:11 UTC

- reply

I burned my account when they sold.

2026-03-23 01:04:25 UTC

- reply

Pushed some pics into differing ethers. Flickr is a travesty, was a real virtual/meatspace community. Some of the best times of the internet.

2026-03-23 00:44:34 UTC

- reply

There is/was a great community there.

2026-03-23 00:35:55 UTC

So who's going to buy out Flickr and republish on Nostr?
I'd tip in.

2026-03-17 12:43:12 UTC

- reply

Yes

2026-03-15 02:58:58 UTC

- reply

Try maple

2026-03-12 18:32:00 UTC

COLDCARD Mk5 https://www.youtube.com/watch?v=n3iQ3UOWgZg

2026-02-21 18:12:33 UTC

Honey badger doesn't care
https://blossom.primal.net/de681eda7ef6cc61a953ab532cb759df16b2522a33eb5cb270329ba42839672d.jpg

2026-02-01 20:34:34 UTC

- reply

Never take in consideration the gheys

2026-01-28 02:22:31 UTC

- reply

Inconvenient

2026-01-22 23:40:19 UTC

- reply

People shouldn't read their encrypted backups on a computer. Keep it air gap

2026-01-20 20:13:42 UTC

https://blossom.primal.net/21274149d27dc82bfde6e1f208d408de2cbc03a6286273d5a511af59e8e6624a.jpg

2026-01-20 02:41:21 UTC

https://blossom.primal.net/780fd80d4cbb43a173dfcf2cb7752e1286f13e69afa10f2817ff06d1f519f9d2.jpg

2026-01-19 13:09:15 UTC

- reply

🤌
#nevent1q…q8ys

2026-01-17 23:21:28 UTC

- reply

💤
#nevent1q…prz4

2026-01-11 17:53:42 UTC

https://blossom.primal.net/dacb03ec817f71e17a7ad1ac5cd2df1aafe36658f51f61f560e1c44e6b6bc1f7.jpg

2026-01-11 17:51:09 UTC

- reply

Many such cases

2026-01-11 17:24:02 UTC

GM is Nikita here?

2026-01-11 16:46:14 UTC

Free Persians breaks a lot of the previous world order.
Changes most of the Middle East.
Obama would not let them have it.
They are taking it.

2026-01-11 16:07:42 UTC

Free Persuians breaks a lot of the previous world order.
Changes most of the Middle East.
Obama would let them have it.
They are taking it.

2026-01-11 01:51:57 UTC

Persia.

2026-01-11 00:44:20 UTC

- reply

Love to see it, fuck the regime.
But this is a poster made from a lady at a parking lot in toronto.
#nevent1q…epdg

2026-01-07 04:45:45 UTC

- reply

🤝 Continuous Innovative Animals
#nevent1q…wdk4

2026-01-04 22:29:44 UTC

https://blossom.primal.net/1a302f75ad539cb9366a909959babc83f80a48cdb294d551e674573723cb614a.jpg

2026-01-04 02:21:29 UTC

🌕
https://blossom.primal.net/d103f66f53db8f362bf595f54ffab2929856eb258295433968ad9bff1e9d55c2.jpg

2026-01-04 00:14:35 UTC

- reply

Me
Too

2026-01-01 14:14:12 UTC

- reply

HNY! Exit.

2025-12-26 21:12:48 UTC

z.ai

2025-12-18 23:52:09 UTC

- reply

Peak internet
#nevent1q…38ga

2025-12-18 18:32:15 UTC

- reply

It's the great quantum bamboozlement, folks want your corn cheap
#nevent1q…44ug

2025-12-14 22:16:34 UTC

Nothing improves marriages like a mud recovery.

2025-12-13 15:18:00 UTC

https://blossom.primal.net/dc47cd8a577bf9eaefdf6b481643af60fc9c0950be04176b1328c5392b805d71.jpg

2025-12-13 15:17:19 UTC

- reply

Raw herring is the best fish, nothing beats it.

2025-12-13 14:44:00 UTC

- reply

These are for a existing bitcoiner

2025-12-13 14:21:36 UTC

Diapers arrived 🎉

2025-12-13 14:18:15 UTC

- reply

He has 4 yonge kids, I doubt alone is part of the mix. 🤣

2025-12-13 12:18:05 UTC

- reply

Ultrathink?

2025-12-12 14:06:52 UTC

📲
https://blossom.primal.net/626b1e6faab1a6cd913bced82ea97209b4b052e119f52ae1572cd4e04c08cc8d.jpg

2025-12-11 18:59:49 UTC

Humanity doesn't move forward untill traditional banks are deprecated.

2025-12-10 19:14:37 UTC

- reply

"What do you mean? TheFacebook.com still around?"

2025-12-09 23:50:47 UTC

If the last mile accepts bitcoin, KYC dies.

2025-12-09 13:02:20 UTC

The internet of shit strikes back
https://blossom.primal.net/3480f7bbdd32d142bf7abca415106f40ffab995233688f0ffa1e49dae99d901f.jpg

2025-12-08 16:44:20 UTC

BBQr testing site https://bbqr.org/js-demo

2025-12-06 15:32:56 UTC

https://blossom.primal.net/a6aed1d58aa3e3d830af6ed3d7703cbb528908c09f51c346557188c42e333861.jpg

2025-12-05 23:06:58 UTC

- reply

Yes
#nevent1q…rgah

2025-12-05 16:52:46 UTC

- reply

Incredibly clever 1 of 8 system.
#nevent1q…zvfq

2025-12-05 12:04:39 UTC

- reply

This timeline is priceless

2025-12-05 01:29:47 UTC

- reply

https://blossom.primal.net/07a063ddb222702f9bd9f0dbacf3978d2d5e44645c5cbf011bed2d08a672d45c.jpg
#nevent1q…s44a

2025-12-03 16:06:35 UTC

- reply

HFSP

2025-12-02 14:52:55 UTC

Today @nprofile…xz5t is preventing me from dopamine hits.

2025-12-01 20:46:52 UTC

- reply

Sucks. But 99% of the users will be on the AppStores.

2025-12-01 17:03:11 UTC

I detest PWAs.
Shiniest experience, if you are not forbidden to do so for your category, man up and App Store it.

2025-11-30 16:49:44 UTC

https://blossom.primal.net/f8dd3b51061244514c6ba68fc023e2ac0394e5a15b7e0d48b518d77dac9f456f.jpg

2025-11-30 16:28:39 UTC

https://blossom.primal.net/ccb87e210f26d47a4d8872761aff376603398e68616ab3f3dd316332fceb0368.jpg

2025-11-30 00:14:22 UTC

https://blossom.primal.net/fbc78fbc499fa3f8f1860139e5c028fe73d945f6967f4085216178acd7a54e3b.jpg

2025-11-29 15:37:23 UTC

There are no amount of features and execution that will bring users from efficient centralized platforms.
Users are not shopping for a washing machine in their slop content consumption.
At some point Nostr will have to hack its way into networks and other things.

2025-11-28 21:32:29 UTC

Peak pepper grinder design https://mannkitchen.com/products/the-original-pepper-cannon-pepper-mill

2025-11-28 01:41:08 UTC

We need a X alternative, we just don't have a fucking clue how we get there.
Doesn't mean we stop trying.

2025-11-27 15:46:59 UTC

Nostr needs hackers https://x.com/0xmitsurii/status/1993942889567764780?s=46

2025-11-25 01:59:01 UTC

CBOR NOSTR 🤤

2025-11-23 16:30:09 UTC

- reply

Practically impossible 🤣🫠
#nevent1q…cayr

2025-11-18 20:16:14 UTC

https://www.tekton.com/blog/jis-vs-phillips-screwdriver-tip-geometry-and-fastener-compatibility

2025-11-14 21:24:29 UTC

Is there a bounty site for 3D printed parts design?
1. Someone creates a bounty w/ initial #bitcoin pledge & design spec
2. Others can pile up & add to the tip pile
3 Someone proposes a claim to the bounty w/ design
4. Bounty creator accept(/deny)
5. Designer gets paid
6. If time expires or bounty creator gives up, all the pledges are released to depositors
DLCs can do this with lightning

2025-11-06 19:46:40 UTC

- reply

We will not have apps or servers with customer information.

2025-11-04 01:37:49 UTC

https://blossom.primal.net/e5bffa676442583805f192683c9c1c28af2b1d282e29639929c047941cdd12eb.jpg

2025-10-17 18:35:16 UTC

- reply

Doesn't mater, the signer is repro

2025-10-17 17:56:47 UTC

- reply

Bull is the best way of buying and selling bitcoin.
#nevent1q…lnvz

2025-10-14 18:12:13 UTC

- reply

I've talked to Antoine in private, it was very productive. Not getting in details, but I think the bridges are being built, the interest in listening to users is there, and hopefully we can all do better.
Time to lower the temperature, everyone have different preferences with a shared rigid system.
Let's all walk on grass a bit 🤙
https://blossom.primal.net/6e97b823eab7cc145b956500bd13e14bed97d051ad167999359642b270eeae87.jpg
#nevent1q…pu2e

2025-10-13 19:49:07 UTC

- reply

How many companies is she running?

2025-10-13 01:13:45 UTC

So much drama over nothing, a real bitcoin business;
$ uptime
9:05PM up 1006 days, 31 mins, 1 user, load averages: 0.25, 0.43, 0.46
$ bitcoind --version
Bitcoin Core version v0.19.1.0-g58ba7c314d552cea8cb024960a8504577aee586f
Copyright (C) 2009-2019 The Bitcoin Core developers

2025-10-12 16:26:58 UTC

https://blossom.primal.net/0af22c49468729e8ee6d968a39c7a75d2e0ae106e78c98c0b9079f96619f3ae2.jpg

2025-09-21 16:59:05 UTC

https://nostr.xxx 💦

2025-09-18 22:51:20 UTC

Both groups feel attacked so both groups act retarded.
Love you all