The Open Standard Collective Context Engine for Humans & Agents. Because Innovation Has No Borders and neither should your AI.
Los Angeles
Joined August 2026
- Tweets3
- Following56
- Followers2
- Likes43
AdnaNetwork retweeted
this paper should be required reading if you use Claude Code.
Wavestone AI Lab analyzed the source code of 11 coding agents, including Claude Code and Codex.
they found the same 7 building blocks inside capable coding agents.
agent = model + harness
here are the 7 parts, from simplest to most advanced:
β 1. Loop: A basic while loop to fully replayable execution runs.
β 2. LLM layer: A single prompt template to multiple provider-specific configurations.
β 3. Tools: Basic bash commands to dozens of typed tools loaded on demand.
β 4. Memory: Raw conversation history to persistent notes that survive across sessions.
β 5. Safety: A simple step limit to rules, reviewer models, and sandboxing.
β 6. Orchestration: Managing one agent or spawning sub-agents to handle parallel tasks.
β 7. Extensions: Skills, MCP, and other ways to extend what the agent can do.
want to read the paper? start with these pages:
π Page 4: The complete 7-part architecture.
π Page 12: Three types of agent loops, including Claude Code and Codex.
π Page 67: 18 design rules for building better harnesses.
π Page 71: A 90-line harness that covers all 7 components.
here's the insight most people miss:
the model gets the attention.
but the harness determines how the agent uses tools, manages context, handles failures, remembers progress, and completes the task.
that's why switching to a smarter model doesn't automatically fix an unreliable agent.
if you want to build production-grade AI agents, learn the harness, not just the model.
I open-sourced Claude Duet!
(same proactive agent, on open models, MIT-licensed)
It's called OpenDuet.
it briefs you every morning, checks your apps while you're away, and only pings you when something needs you.
Anything that sends, posts, or pays waits for your approval.
Each bot gets its own open model and a plain MEMORY.md you can edit.
built with @mastra, @nebiustf and @composio
Repo: github.com/Studio1-OSS/OpenDβ¦
AdnaNetwork retweeted
Delighted to share a landmark preprint from our lab, using genome language models and AI agents for biological discovery and genome mining!
The stretches of DNA between protein-coding genes are often called the genomeβs βdark matter,β but they are far from empty. They encode RNA molecules that sense metabolites, regulate genes, and control some of molecular biology's most important tools (such as programmable genome editors e.g. CRISPR/bridge)
Much of this biology is unexplored. One reason is that RNAs can retain similar structures and functions even as their DNA sequences change dramatically, making their relationships difficult to recognize with conventional computational tools