Agentic Engineering Orchestration, context, and tool use for agents built to work, not demo. https://iggym.github.io/agentic-engineering/ 2026-10-06T00:00:00Z Agentic Engineering The Retry Echo https://iggym.github.io/agentic-engineering/articles/the-retry-echo.html 2026-10-06T00:00:00Z Your agent didn't double-charge the customer because it reasoned badly. It retried a write it had no way to observe. Description Drift https://iggym.github.io/agentic-engineering/articles/description-drift.html 2026-10-06T00:00:00Z Your agent didn't forget how to use the tool. The tool changed what it tells your agent, and nobody diffed it. Instruction Sediment https://iggym.github.io/agentic-engineering/articles/instruction-sediment.html 2026-10-06T00:00:00Z Your agent isn't ignoring your rules. It's obeying all of them, including the ones whose reason left the team months ago. The Paper Plan https://iggym.github.io/agentic-engineering/articles/the-paper-plan.html 2026-10-06T00:00:00Z Your agent's plan looked right because nothing ever held execution to it. The plan is a document; the trajectory is what ran. Handoff Fiction https://iggym.github.io/agentic-engineering/articles/handoff-fiction.html 2026-10-06T00:00:00Z Your reviewers didn't miss the bug. They reviewed the agent's description of the change, and the description was fiction. Misdirection in the Tool https://iggym.github.io/agentic-engineering/articles/misdirection-in-the-tool.html 2026-08-13T00:00:00Z Your tool isn't broken — your retrieval is lying to it. Diagnosing Verify Amnesia https://iggym.github.io/agentic-engineering/articles/diagnosing-verify-amnesia.html 2026-05-15T00:00:00Z The failure didn't originate at the tool call; it originated at the Verify stage, which evaluated the 4,800th attempt in a vacuum. Divide And Debug https://iggym.github.io/agentic-engineering/articles/divide-and-debug.html 2026-04-30T00:00:00Z Multi-agent orchestration isn't a capability upgrade; it is a debugging upgrade wearing a performance costume. The Harness Mirage: Why Your Agent Infrastructure Is Lying to You https://iggym.github.io/agentic-engineering/articles/the-harness-mirage.html 2026-04-16T00:00:00Z Most agent failures aren't model reasoning errors—they're the harness mirage: unconstrained execution loops passing dirty state through shallow verification gates. Stop Fixing Tool Syntax https://iggym.github.io/agentic-engineering/articles/stop-fixing-tool-syntax.html 2026-03-18T00:00:00Z Your agent rarely breaks at tool execution; it breaks two layers upstream when context truncation feeds it half a truth. Phantom Inheritance: When Your Subagents Remember What You Told Them To Forget https://iggym.github.io/agentic-engineering/articles/phantom-inheritance-subagents-remember.html 2026-03-14T00:00:00Z Isolation isn't about what you don't pass in the prompt — it's about what your runtime smuggles back in the tool results. When One Subagent's Memory Becomes Everyone's Problem https://iggym.github.io/agentic-engineering/articles/when-one-subagents-memory-becomes-everyones-problem.html 2026-03-12T00:00:00Z Isolation doesn't make agents smarter. It stops one bad memory from becoming a system-wide lie. The Side-Effect Slipstream: Where Agent Runtime Risk Actually Lives https://iggym.github.io/agentic-engineering/articles/the-side-effect-slipstream.html 2026-02-19T00:00:00Z Stop red-teaming prompts while ignoring unisolated runtimes. Most agent exploits leverage the side-effect slipstream: unmonitored state mutations passing through surface-level tool verification. The Verify Gap https://iggym.github.io/agentic-engineering/articles/verify-gap.html 2026-02-13T00:00:00Z The check that should catch the failure usually runs – but it looks at the wrong layer. Context Bleed https://iggym.github.io/agentic-engineering/articles/context-bleed.html 2026-02-13T00:00:00Z Your context isn't isolated — it's a shared wound that infects every subagent. Moat Trap https://iggym.github.io/agentic-engineering/articles/moat-trap.html 2026-02-13T00:00:00Z The model is a commodity — your harness is the only durable asset. Verify Theater https://iggym.github.io/agentic-engineering/articles/verify-theater.html 2026-02-13T00:00:00Z Your evals pass, your agents fail — because you're checking answers, not trajectories. Diagnosing the Phantom Argument https://iggym.github.io/agentic-engineering/articles/diagnosing-the-phantom-argument.html 2026-02-12T00:00:00Z Your tool call didn't fail because the model is dumb. It failed because of a Phantom Argument—a parameter ghostwritten by a 3-day-old RAG index. The Green-Light Lie: Your Agent Passed Eval and Failed the Job https://iggym.github.io/agentic-engineering/articles/green-light-lie-eval-failure.html 2026-02-10T00:00:00Z We grade agents like chatbots — final answer correct — while trajectory failures, harness variance, and flawed verifiers make leaderboards lie. Never Committed https://iggym.github.io/agentic-engineering/articles/never-committed.html 2025-04-15T00:00:00Z Your agent didn't run out of context. Your state layer never committed. Catching Fire https://iggym.github.io/agentic-engineering/articles/catching-fire.html 2025-03-14T00:00:00Z Your shared-context architecture isn't making your agents collaborative — it's making them contagious. The Final Answer Trap https://iggym.github.io/agentic-engineering/articles/final-answer-trap.html 2024-04-25T00:00:00Z A passing final answer is the worst possible signal of agent capability.