Built an offline-verifiable receipt layer for agent/tool actions ... looking for

<!-- SC_OFF --><div class="md"><p>I've been following the IETF drafts for agent identity, and I realized there's a massive hole: User-to-Operator trust. We have ways for services

<!-- SC_OFF --><div class="md"><p>I've been thinking about this as local models get better at </p> <p>tool use — shell access, file system, HTTP requests, MCP. </p> <p>If a prompt

<!-- SC_OFF --><div class="md"><p>I've been thinking about why local-first AI agent architectures are getting serious enterprise traction in 2026, beyond the obvious &quot;keep

<!-- SC_OFF --><div class="md"><p>I've been working on MDA (Modular Dynamic Architecture), an online associative memory system for LLMs. Here's what I learned building it.</p>

<h2> The scenario </h2> <p>Your AI agent just deleted a customer record. Three months later, an auditor asks you to prove:</p> <ul> <li> <strong>What tool actually ran</strong>

<!-- SC_OFF --><div class="md"><p>we run an agent thing in production and we use langfuse for traces.</p> <p>last month our agent started refusing requests it should have

<!-- SC_OFF --><div class="md"><p>I have built a programmable governance layer for AI agents. I am considering to open source completely. Looking for feedback.</p> <p>Agent demos

<!-- SC_OFF --><div class="md"><p>Been heads down building graphify since April 5th. The idea: instead of stuffing your entire codebase into context every session, build a