Seonglae ChoGenerative Agent
- Project Folder (Repository) = System of Record: All knowledge must exist explicitly within the repo.
AGENTS.mdshould be kept short, with actual knowledge distributed indocs/structure.- Documents are also mechanically verified with CI and lint.
- Agent Legibility First: Code style prioritizes structures that agents can easily understand over human preferences.
- Architecture layer structure: Types → Config → Repo → Service → Runtime → UI
- Dependency direction enforced by lint. Rules are enforced by code, not documentation.
- Agents directly handle UI, logs, and metrics
- Chrome DevTools Protocol integration.
- LogQL / PromQL access available, feedback loop must be automated
How can we scale this work from sequential to parallel is going to be the next jump
- planning AI scheming
- memory
- tool interface
Specifically
- Tool learning
- System Calls
- Socket
- Headless Browser access to pretend human
- CLI access
- Curl
- GET - Search
- POST - Communication
- Permanent Storage like vector database
- Context Selection based on vector similarity
- Long-term memory by weight update
- Vanishing to prevent manage capacity based on LLM decision or dynamic sparsity
- Self-replication, Self-restart
- Self-initialization prompt, Self-editing
- Asynchronous communication port
- Interrupt
- …
Text to Task Generative Model
- Photoshop like AI software
AI Agent Notion
AI Agent Usages
Tips (AI Service, AI Agent)
- Split into micro-agents with single roles. (Model-based agents still don't perform well due to insufficient data accumulation)
- Keep only frequently used tools. (terminal, Language Server Protocol)
AI Agency
Decomposing Agency — capabilities without desires — LessWrong
People sometimes say that AGI will be like a second species; sometimes like electricity. Both may have elements of truth. We need concepts which let us think clearly about that region in-between the two.
https://www.lesswrong.com/posts/jpGHShgevmmTqXHy5/decomposing-agency-capabilities-without-desires
Google whitepaper
Agents
Authors: Julia Wiesinger, Patrick Marlow and Vladimir Vuskovic
https://www.kaggle.com/whitepaper-agents
Workflow orchestration (Prompting Optimization)
Agent Recipes
Explore common agent recipes with ready to copy code to improve your LLM applications.
https://www.agentrecipes.com/
Workflow-based vs. Model-based
In the Matter of OpenAI vs LangGraph
The silent war in Agent Engineering gets loud.
https://www.latent.space/p/oai-v-langgraph
Agent Software
- Parity: Agents should be able to perform any action through tools that is possible through the UI.
- Granularity: Tools should be atomic primitives without judgment. The agent makes the decisions.
- Composability: New features are composed through prompts alone without code changes.
- Emergent Capability: Agents should be able to solve unplanned requests by composing available tools.
- Improvement over TimePerformance: improves without deployment through context accumulation and prompt refinement alone.
Agent-native Architectures: How to Build Apps After Code Ends
A new software paradigm for the AI age
https://every.to/guides/agent-native
