01 / FLAGSHIP
Frontend Design
Art direction, implementation, critique, responsive QA, and deployment handoff for interfaces that need a coherent visual world.
Surface → Design Read → Proof → Build → Visual QA
↗
Multidisciplinary Designer & AI Workflow Architect
Design systems.
Agent workflows.
Proof that ships.
I am TIKAZ — a designer across brand, graphic, product, and frontend disciplines who also writes code. I build open Codex workflows that turn creative direction and technical execution into traceable systems.
Context proof69.73% smaller on six long-context tasks · declared facts and anchors retained · not a cross-project leaderboardACTIVE SYSTEMTIKAZ AI Skills for Codex
01Frontendart direction
02Videosource evidence
03Contextmeasured efficiency
04Engineeringverified delivery
05Researchtraceable decisions
06Presentationrendered artifacts
07Visualpublishing QA
Canonical source30 Skillsv0.5.0
System architecture
One source of truth.
Seven focused entry points.
The monorepo owns architecture, provenance, validation, and cross-suite handoffs. Seven distribution repositories are generated from it so each workflow can be discovered and installed independently without creating eight drifting codebases.
- 01One suite owns the deliverable.
Specialists support the outcome without competing for control.
- 02Evidence travels with the work.
Sources, confidence, licenses, and unresolved gaps survive each handoff.
- 03Objective gates close the loop.
Rendered output, tests, reviews, and file evidence outrank agent confidence.
The collection
Seven workflows, designed to compose.
02 / FLAGSHIP
Video Intelligence
Cross-platform video reading that keeps metadata, transcript, keyframe, verification, and inference at distinct evidence levels.
Sources → ASR → Keyframes → Cards → Synthesis
↗
03 / NEW
Context Economy
Hard-budget context packs, recoverable checkpoints, six-dimension audits, and public benchmarks that keep savings and fidelity separate.
Pack → Checkpoint → Audit → Benchmark
↗
04
Engineering
Specification, code intelligence, implementation, tests, review, security, and release evidence.
Scope → Impact → Code → Test → Release
↗
05
Knowledge & Research
Source normalization, evidence ledgers, disagreement, confidence, product decisions, and reusable personal memory.
Question → Sources → Evidence → Decision
↗
06
Presentation
Narrative-first HTML, editable PowerPoint, and editorial web decks with page contracts and rendered QA.
Narrative → Format → Pages → Render → Artifact
↗
07
Visual Content
Provider-neutral illustration direction, concise writing, lawful music discovery, accessibility, and publishing QA.
Thesis → Style → Shot → Create → Verify
↗
Method demonstrations
Show the method.
Do not invent the outcome.
DEMONSTRATION 00
Efficiency with visible limits
Six tasks over one synthetic long source shrink 4,698 estimated tokens to 1,422 while preserving declared facts and expected anchors. The 30 micro cases grow because protocol overhead dominates, so the two profiles remain separate.
Read the Context Economy proof ↗DEMONSTRATION 01
Proof before scale
A dense operations interface is classified as an Operate surface. The hardest representative desktop/tablet slice must pass hierarchy, scan path, and action clarity before the system expands.
Read the design proof ↗DEMONSTRATION 02
Evidence is not interchangeable
A transcript claim, a visual observation, and an external verification keep distinct timestamps and confidence. The synthesis cannot erase what remains unknown.
Read the evidence card ↗About TIKAZ
Design depth.
Technical range.
Systems thinking.
I work across brand design, graphic design, product design, and frontend design — and I understand the code that turns decisions into real interfaces. I use Codex and other OpenAI tools as working instruments, then shape their output with design direction, engineering constraints, and verifiable evidence.
Brand systemsGraphic designProduct designFrontend designCode literacyAgent workflows
Propose a workflow or report an issue