Make AI Explanations Easier to Read with ASD-STE100
Use Simplified Technical English to make AI explanations clearer: short sentences, consistent names, explicit actions, and a prompt you can copy.
Read articleTopic Pillar & Hub
Agent Workflows
Architecture, Evaluation Loops, and Context Control for Autonomous Coding Agents
Moving beyond toy demos requires robust engineering around coding agent loops. This topic covers context governance, verification harnesses, MCP integration, and the failure modes encountered when agents write production code.
Core Engineering Pillars
Loop & Graph Engineering
Designing state machines and decision trees that guide agents from requirement to verified PR.
Context Governance
Keeping active working memory clean and isolating agent execution from noisy logs.
Verification Harnesses
Automated test execution, lint checks, and self-healing review loops.
Recommended Progression
- 1. Establish Execution Contracts
Define crisp acceptance criteria before handing control to autonomous loops.
- 2. Set Up Verification Feedback
Ensure the agent receives compiler/test errors directly in its feedback loop.
- 3. Implement Guardrails
Prevent destructive actions and infinite retry loops with deterministic exit conditions.
Frequently Asked Questions
What is the main bottleneck in autonomous coding agents?
Context pollution and lack of deterministic verification. Agents rarely fail on simple logic; they fail when error logs flood the context window.
Field Reports & Case Studies
Agent Workflows Articles (16)
I Open-Sourced a PDF Invoice Skill for Coding Agents
I open-sourced a PDF invoice skill for coding agents. One prompt updates days, discount, fees, and the balance due on a generated PDF.
Read article
Stop Reviewing Agent Code: Lauren Tan's System
Lauren Tan merged 1,000 PRs last month with agent auto-merge. Her 4-layer system: verification, feature maps, CI constraints, and guardrail accumulation.
Read article
Inside DeepSeek Harness and Its Plugin-Based Agent Runtime
DeepSeek Harness separates models, tools, sessions, permissions, and interfaces into plugins that builders can replace and combine.
Read article
LoopX: A Local Control Plane for Long-Running Agent Work
LoopX keeps goals, evidence, ownership, quotas, and handoffs stable across long-running agent sessions without replacing the agent runtime.
Read articleCan Kimi K3 Handle a Real Production Backend?
A hands-on Kimi K3 backend audit across 68K lines of FastAPI code, including query fixes, measured results, and the full task cost.
Read articleFrom Loop to Graph Engineer: The 5-Layer Stack
Prompt, context, harness, loop, graph: five stacked layers that move reliability out of the model and into the system around it.
Read articleAI Is Turning Everyone Into a DevOps Engineer
A DigitalOcean migration that used to take half a day. I set up SSH, described the target state, and Codex did the rest.
Read articleGPT-5.6 Sol Wiped a Developer's Mac. Protect Yours.
On July 10, GPT-5.6 Sol deleted a tester's home directory after a failed $HOME expansion. How the incident happened, and three layers of defense.
Read articleHeadroom: Context Governance for Coding Agents
Headroom sits between your agent and the model, classifying and compressing tool output before it fills the context window.
Read articleAgents can now deploy to Cloudflare without signing up
Wrangler --temporary gives an agent a 60-minute Cloudflare account, a live workers.dev URL, and a claim link. No OAuth, no MFA, no human click.
Read articleDoubao's Agent Architecture: 25 Skills, 283 Files
A leaked Doubao skill pack shows 25 modular skills and 283 files organized around real work objects, not one giant agent.
Read article40 agent loops you can copy into Claude Code or Cursor
loops.elorm.xyz packages 40 agent loop templates by category so you can copy a kickoff prompt into Claude Code, Cursor, Codex, or Gemini CLI.
Read articleLoop Engineering: 14-Step Agent Builder Roadmap
A practical roadmap for deciding whether to build an agent loop, then adding automation, state, verification, tools, and security controls in the right order.
Read articleLoop Engineering: The Next Layer After Prompt Engineering
Peter Steinberger and Boris Cherny both say the job is designing loops, not writing prompts. Feedback, stop conditions, and token cost decide if it works.
Read articleHow to Build a Self-Improving Company with AI
YC partner Tom Blomfield's framing: make company knowledge machine-readable, then run loops that sense, decide, act, and improve overnight.
Read article