Engineering journalVol. 03 · Agentic AI55 essays · new weekly

Field notes from the frontier of agentic systems

Deep dives on large language models and cognitive architectures — paired with live, runnable playgrounds so every idea can be executed, not just read.

Deep-dive articles: 55.; Deep-dive articles
Execution modes: 6.; Execution modes
Live agent labs: 4.; Live agent labs

The writing

Latest deep dives

Multi-AgentSeptember 4, 2026

ADF Coder

The Autonomous Dev Framework has hardened into a governed protocol (ADP v1.2): a versioned source knowledge graph, a 7-rule task lifecycle with echo checkpoints, and validation gates that make self-assessed confidence irrelevant.

Read essay

FrameworksAugust 26, 2026

Sandbox Architectures for Untrusted Code Execution

Containers, microVMs, WebAssembly: escape resistance versus cold starts versus capability. Ephemeral sandboxes, proxied egress and why sandbox depth is a product feature.

Read essay

FrameworksAugust 19, 2026

The Confused Deputy Problem in Tool-Using Agents

Ambient authority is why one successful injection inherits everything. Capability tokens, per-task identities and narrowing delegation chains — the 1988 cure, applied.

Read essay

KnowledgeAugust 12, 2026

Prompt Injection: The SQL Injection of LLM Systems

Direct is manageable; indirect — the poisoned page, the crafted ticket — is the nightmare. The honest defense stack, and why architecture inherits what attention cannot hold.

Read essay

KnowledgeAugust 5, 2026

Evals-Driven Development

Write the evaluation first, build the smallest system that passes: red-green-refactor for AI features, from smoke suites to nightly datasets that catch slow drift.

Read essay

KnowledgeJuly 29, 2026

LLM-as-Judge: Strengths and Blind Spots

Position bias, verbosity bias, self-preference: the documented failure modes of model-graded evaluation — and the calibration stack that makes the proxy trustworthy.

Read essay

FrameworksJuly 22, 2026

The Outbox Pattern for Reliable Agent Side-Effects

State change plus event publication in one transaction, relayed at-least-once: deleting the entire category of "the run succeeded but nobody was told."

Read essay

FrameworksJuly 15, 2026

Backpressure in Streaming Agent Pipelines

Bounded queues, value-classed shedding and pauseable generation — propagating "slow down" upstream instead of buffering toward a 3 a.m. OOM.

Read essay

FrameworksJuly 8, 2026

CQRS for Agent State

One append-only run log, a dozen purpose-built projections: how command-query segregation lets UIs, auditors and finance teams read the same truth without fighting over its shape.

Read essay

Don't just read — run it

Every concept here has a live execution mode.

Bring your own Gemini key and watch agentic loops, meta-cognitive workflows and ADF builds trace themselves in real time.