I counted the pastes last Tuesday. Forty-one, before lunch. Same context block, same brand guidelines, into a new chat each time. The fix wasn't a smarter prompt, it was pasting the context once per project instead of once per question.

#contextengineering #promptengineering

New arXiv paper frames the context files that LLM agents load each session as a capacitated assortment problem, showing that appending every candidate instruction can be arbitrarily worse than picking an optimal subset…

#LLMAgents #ContextEngineering #AIResearch
https://arxiv.org/abs/2610.11007

Hire someone brilliant, give them zero context about your customers, your voice, what good looks like. You'd call that bad management. That's exactly how most people use AI, then they're shocked the output is generic.

#ai #contextengineering

The update record checked more than the new files: it confirmed the old filenames were gone. If both generations remained, "use the current version" could still leave a choice. I want an update report to show what the next session can find. #AI #ContextEngineering

A new hire gets a brief before their first task: who we are, who we sell to, how we sound. Your AI gets a one-line request and somehow has to guess all of that. Write the brief once. Paste it in before every request. The gap closes fast.

#contextengineering #promptengineering

One design note had to reconcile an older "not applied" label with a later placement check. It used the later check for the current baseline while leaving the next candidate pending. One status label could not describe both. #AI #ContextEngineering

The tests passed, but my everyday AI setup still hadn't been switched over. Both were in the same execution report. Reading only "passed" would have made unfinished work look complete. The useful detail was what remained undone. #AI #ContextEngineering

You'd never hand a new hire one line and expect great work. No context on the company, the reader, the standards. That's exactly what we do to AI, then call the output generic. It's not undertrained. It's un-onboarded.

#contextengineering #promptengineering

A new arXiv paper proposes structured memory that decouples stored experience from active context for token-efficient continual learning in LLM agents. This could make long-running enterprise or scientific…

#LLMs #ContextEngineering #AIagents #MachineLearning
https://arxiv.org/abs/2610.02687

Without tests, you’re just guessing.

Registries don’t keep your codebase safe. Testing does.

Patrick Debois on testing agent skills, context & error budgets for reliable AI deployments.

🔗 Watch now: buff.ly/kIAuTeo

#InfoQ #AIAgents #SoftwareTesting #ContextEngineering

Context Engineering in Production #20

Don't measure AI coding only by output quality.

Track:

• context size
• prompts/task
• retrieval accuracy
• cost/task
• time/task

If context engineering works, those numbers should move.

#ContextEngineering #AIEngineering #SigMap

Context Engineering in Production #19

"AI hallucinated."

Don't stop there.

Check:

Fake symbol?
Fake file?
Invented API?
Wrong parameter?
Unsupported claim?

Groundedness can be tested against repository evidence.

#ContextEngineering #AIAgents #SigMap

Context Engineering in Production #18

The AI produced an answer.

Now verify it.

Do those files exist?
Are those symbols real?
Are the parameters correct?
Is the answer grounded in retrieved context?

Generate → Verify.

#ContextEngineering #AIAgents #SigMap

Context Engineering in Production #17

Changed 4 files?

Don't rebuild context from 4,000.

Start with the diff.

Then expand outward only when dependencies require it.

Context should grow with evidence, not repository size.

#ContextEngineering #Git #AIAgents #SigMap

Context Engineering in Production #16

Before changing code, ask:

Who depends on this?

Finding the target is only half the job.

Understanding the blast radius tells you what the change could break.

Navigate → Impact → Edit.

#ContextEngineering #BlastRadius #SigMap

Context Engineering in Production #15

Traditional workflow:

Search → open → wrong file → retry.

Better workflow:

Question → rank relevant code → inspect only what matters.

Exploration has a token cost.

Make retrieval earn every file it opens.

#ContextEngineering #AgentNavigation #SigMap

Context Engineering in Production #14

Context compression shouldn't mean:

"Remove useful information."

It should mean:

"Represent useful information with fewer tokens."

Signatures, relationships and ranked context preserve signal while dropping noise

#ContextEngineering #TokenEconomics #SigMap

Context Engineering in Production #13

Don't go:

Repository → AI

Go:

Repository
↓
Module
↓
File
↓
Symbol
↓
Implementation

Each step removes irrelevant context.

By the time AI sees code, most noise should already be gone.

#ContextEngineering #AIAgents #SigMap

Context Engineering in Production #12

Finding "RankingEngine.score()" is step one.

Now ask:

Who calls it?
What does it call?
What depends on it?

Code becomes useful context when you understand its relationships.

#ContextEngineering #CodeRelationships #SigMap

Context Engineering in Production #11

Don't start by loading implementation.

Start with structure:

What exists?
Where?
Who calls it?
What does it call?

Then open implementation only where needed.

Structure first.
Details on demand.

#ContextEngineering #RepoIntelligence #SigMap