Adversarial view
Attack surfaces, trust boundaries, defensive controls, failure paths, and the evidence required to support a finding.
QUESTION: WHAT CAN BE ABUSED?FIELD NOTES / RESEARCH LOG
DOC DY-BLOG / RESEARCH LOGFIELD
NOTES.
Practical research, engineering breakdowns, and observations from work across security and intelligent systems.
02 / RESEARCH COMPASS
Four lenses for reading the archive.
The compass describes the questions behind the writing. It does not imply a published note exists for every topic; the archive below remains the source of truth.
Structural view
Interfaces, state, dependencies, deployment conditions, and the contracts that keep behavior understandable.
QUESTION: HOW DOES IT HOLD TOGETHER?Evaluation view
Grounding, retrieval, data quality, refusal boundaries, human review, and where model assistance is genuinely useful.
QUESTION: WHAT MAKES THE OUTPUT TRUSTWORTHY?Field view
Observability, release evidence, incident learning, recovery paths, and the gap between a passing build and a working service.
QUESTION: WHAT HAPPENS AFTER RELEASE?Strong notes often cross more than one track: a security decision changes architecture, an AI evaluation changes product behavior, and an operational failure changes both.
02 / FEATURED
Start here.
A short selection of foundational notes and current investigations.
How a Flask portfolio platform separates public content from administration
A field note on separating public pages, mutable content, contact input, and administration in a Flask portfolio application.
Read note ↗What makes a prototype project page trustworthy?
A practical evidence model for describing prototype systems without turning an implementation claim into a production claim.
Read note ↗03 / ARCHIVE
All writing.
Ordered from the latest dispatch backward.
How a Flask portfolio platform separates public content from administration
What makes a prototype project page trustworthy?
04 / PUBLISHING STANDARD
What a useful field note should contain.
Writing is treated as part of the verification process: a way to expose reasoning, supporting material, uncertainty, and the decision that follows.
01QUESTIONDefine the decision or failure being investigated.
02EVIDENCEShow observations, sources, tests, or reproducible behavior.
03LIMITSState what the material cannot establish.
04OPERATIONAL TAKEAWAYTranslate the finding into a decision, control, or next test.