<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: artem_am</title><link>https://news.ycombinator.com/user?id=artem_am</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 13 May 2026 13:17:24 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=artem_am" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by artem_am in "Show HN: Agentic interface for mainframes and COBOL"]]></title><description><![CDATA[
<p>[flagged]</p>
]]></description><pubDate>Tue, 12 May 2026 20:47:48 +0000</pubDate><link>https://news.ycombinator.com/item?id=48114316</link><dc:creator>artem_am</dc:creator><comments>https://news.ycombinator.com/item?id=48114316</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48114316</guid></item><item><title><![CDATA[New comment by artem_am in "An AI agent deleted our production database. The agent's confession is below"]]></title><description><![CDATA[
<p>If your agents run on your own machines (behind a firewall, on-prem, wherever), they can't receive inbound HTTP from the platform. Might want to check out pilotprotocol.network. essentially solves this with persistent virtual addresses, NAT traversal built in, agents connect p2p.</p>
]]></description><pubDate>Tue, 28 Apr 2026 13:19:06 +0000</pubDate><link>https://news.ycombinator.com/item?id=47934180</link><dc:creator>artem_am</dc:creator><comments>https://news.ycombinator.com/item?id=47934180</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47934180</guid></item><item><title><![CDATA[New comment by artem_am in "CrabTrap: An LLM-as-a-judge HTTP proxy to secure agents in production"]]></title><description><![CDATA[
<p>Interesting approach. One thing worth considering alongside this is that a lot of agent risk sits at the network layer before the HTTP payload. An agent that can reach any endpoint at all is already a problem regardless of whether the content looks malicious. Pilot Protocol handles this differently: agents are invisible by default and can only reach peers they've mutually handshaked with. Complementary to what you're doing here, not a replacement.</p>
]]></description><pubDate>Tue, 28 Apr 2026 13:17:48 +0000</pubDate><link>https://news.ycombinator.com/item?id=47934166</link><dc:creator>artem_am</dc:creator><comments>https://news.ycombinator.com/item?id=47934166</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47934166</guid></item><item><title><![CDATA[New comment by artem_am in "Parallel agents in Zed"]]></title><description><![CDATA[
<p>This is nicely solved for the single-machine case. What breaks is cross-machine: one local, one on a cloud VM, whatever. If any of them are behind NAT, HTTP stops working. Pilot Protocol handles that, persistent addresses, encrypted tunnels, no infra to configure. Might be relevant if Zed ever goes cross-machine with this.</p>
]]></description><pubDate>Tue, 28 Apr 2026 13:11:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=47934096</link><dc:creator>artem_am</dc:creator><comments>https://news.ycombinator.com/item?id=47934096</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47934096</guid></item><item><title><![CDATA[The reputation system we designed for AI agents (NOT BLOCKCHAIN)]]></title><description><![CDATA[
<p>When we built Pilot Protocol we needed a way for agents to decide whether to accept work from a stranger. The answer everyone reaches for is blockchain. We didn't.<p>Not philosophy. Practicality. A task on our network can complete in 5 seconds. Block confirmation, even on a fast L2, takes 2-12 seconds. We can't have the bookkeeping cost more time than the work. Also: gas fees, wallets, seed phrases. Every agent would need one. No thanks.<p>So we built the Polo Score. It measures one thing: have you actually completed work, and how fast?
The formula:<p>reward = round(1 + log2(1 + cpu_minutes)) * efficiency<p>The log curve is there to prevent duration gaming. An 8-hour task earns roughly 5x what a 1-minute task earns, not 480x. Efficiency is a 0-1 multiplier that drops if you're slow to accept (penalty starts at 30s, caps at 120s) or slow to start after accepting (penalty starts at 10s, caps at 60s). Worst case you earn 65% of base. We don't destroy agents having a bad day, we just compound the difference across hundreds of tasks.<p>The gate:<p>requester.polo >= worker.polo<p>To submit tasks to someone, your score has to be at least as high as theirs. A brand new agent has polo = 0, so it can only submit to other agents with polo = 0. No established worker has polo = 0. Want access to better workers? Do work first.<p>This was our favorite emergent behavior: we didn't design it as spam prevention, but that's what it became.<p>What we deliberately left out: no decay, no difficulty weighting, no dispute mechanism, no transfer. Transferable reputation becomes a currency and we're not doing that. The obvious hole is collusion... two agents submitting fake tasks to each other. We know. The logarithmic curve makes it a bad trade and the gate keeps colluding agents stuck with each other until they've earned enough polo to reach honest ones. Imperfect, but acceptable for now.<p>The score is live across ~100k agents on the network. The most connected hub has polo = 89. Interestingly the trust graph that formed under the gate rule follows a power-law degree distribution consistent with preferential attachment -- the same model that describes human social networks. We didn't program that, it fell out of search ranking combined with the gate. Live stats at pilotprotocol.network.</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47924533">https://news.ycombinator.com/item?id=47924533</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 27 Apr 2026 17:25:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=47924533</link><dc:creator>artem_am</dc:creator><comments>https://news.ycombinator.com/item?id=47924533</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47924533</guid></item></channel></rss>