DeepResume
AI Applications
A CV-to-offer workspace connecting resume evidence, target JDs, learning plans, and career proof.
Next.jsTypeScriptFastAPIPython
EVIDENCE-LEDLIMITS SHOWN
01 PERSONAL SYSTEM / 2026
wujue / Dithob
A programmer focused on AI applications, agent tooling, and developer efficiency — also an ACG enthusiast who enjoys running, fitness, and CTRL. I turn vague ideas into working systems and document projects, notes, and verification here. Open to directions beyond any single role.
01Open sourceSOURCE
02Evidence-ledEVIDENCE
03Limits shownLIMITS
04Always buildingBUILD
02 / SELECTED WORK
Selected projects
Projects with source links, verification evidence, and explicit limitations.
PaperQAAgent
Paper Research & RAG
A local-first paper research workspace for PDF reading, page-linked retrieval, and citation-grounded answers.
PythonFastAPINext.jsPyMuPDF
EVIDENCE-LEDLIMITS SHOWN
Youcansee
Agent Skill / OCR
A multi-tier image understanding and OCR skill for coding agents, with fallback candidates and safe configuration boundaries.
PythonOCRAgent SkillOpenAI-compatible API
EVIDENCE-LEDLIMITS SHOWN
media-content-distiller
Developer Tooling
A subtitle-first media distillation tool separating one-off URL work from reusable CLI workflows.
Node.jsCLIHTTP APIMarkdown
EVIDENCE-LEDLIMITS SHOWN
03 / CAPABILITIES
What I work on
Capability groups backed by public project evidence, not a badge wall.
AI Applications
RAG / PDF / Paper / Resume / Agent
Agent & Developer Tools
Skills / CLI / OCR / Media / Automation
Engineering Quality
Testing / Validation / CI / Observability / Safety
04 / CURRENT FOCUS
What I am building now
- 01PaperQAAgentPage-linked evidence and local-first paper research.
- 02DeepResumeA loop connecting resumes, JDs, learning plans, and career evidence.
- 03Agent toolingInstallable, verifiable skills and CLI workflows.
05 / WORKFLOW
My workflow
From facts to systems, then to evidence you can inspect.
Verify first
Verify claims against repos, issues, and run logs; label conclusions by evidence.
Shape the system
Turn repeated work into reusable skills, CLIs, and scripts with reviewable artifacts.
State the boundary
Ship test evidence (pytest, Golden Datasets), known limits, and next steps.
06 / LATEST NOTES
Latest notes
Recently published notes with sources, evidence, and open questions.
From manual codex relay config to automated model provisioning: the birth of a pi skill
From a real need (wiring a Codex relay station into pi) through per-model probing, skill generalization, npm publishing, and pi.dev listing — with six reproducible pitfalls including npm 2FA and marketplace index lag.
A practical manual for Agent Skills
A practical beginner manual covering Agent Skill concepts, structure, setup, authoring, debugging, and reuse.
Git and GitHub core concepts
A structured guide to Git version control, GitHub collaboration, branches, commits, merges, and practical workflows.
Pi Agent engineering guide
An engineering walkthrough of Pi, from its npm monorepo and model abstraction to the agent loop and session storage.
07 / NOTES
Article index
Keeping the reasoning, validation, and boundaries behind the projects.