Engineering Papers

A complete engineering methodology for teams building with AI coding agents. The series opens with a sixteen-mode failure taxonomy organized into four priority tiers, then constructs the structural defenses that address each one: workflow architecture, sub-agent design patterns, checkpoint verification, provenance chains, goal-fidelity checks, and a human review runbook calibrated to risk level.

• 01

  Failure Modes and Engineering Safeguards

  The complete sixteen-mode failure taxonomy across four priority tiers, LLM-general vs. agentic-specific classification, and a roadmap for the series. Explains why the most dangerous failures are the ones that produce plausibly correct output.

• 02

  Security Review Workflow — Part 1

  Introduces the five domain-independent structural principles through a concrete security review use case. Establishes the self-scaffolding checklist pattern that grounds the entire methodology.

• 03

  Security Review Workflow — Part 2

  Defines the seven sub-agent archetypes—Researcher, Analyst, Validator, Executor, Auditor, Synthesizer, Coordinator—with enforced tool restrictions and handoff contracts.

• 04

  Security Review Workflow — Part 3

  Automated remediation with structured human checkpoints: how to gate on risk tier, preserve audit trail, and keep human review focused on judgment calls rather than mechanical verification.

• 05

  Security Review Workflow — Part 4

  Live demonstration of the full workflow end-to-end, with annotated outputs showing what correct agent behavior looks like at each stage and how failure modes manifest in practice.

• 06

  Security Review Workflow — Part 5

  Generalizing the security review pattern to testing review, API review, migration review, and performance auditing—showing which elements are domain-specific and which transfer unchanged.

• 07

  Beyond Security — Generalizing the Workflow

  Extracts the domain-independent orchestration pattern and applies it to automated test generation, API code review, database migration review, and performance auditing with complete sub-agent definitions.

• 08

  Advanced Mitigations and Structural Remediations

  Detailed implementation of remediations across all four failure tiers: checkpoint-verifier sub-agent, provenance chains, re-grounding gates, goal-fidelity checks, scope boundary enforcement, consistency reconciliation, and irreversibility gates.

• 09

  Human Review Runbook

  The human side of the methodology: a structured review process for evaluating agent output, calibrated to the four priority tiers. Where to apply scrutiny, what to spot-check, and when to reject versus remediate.