An open-source Codex Skill for adaptive, evidence-based mastery of software engineering interview topics.
中文使用指南 · Architecture · Development · Contributing
Super Study turns Codex into a progressive learning coach for frontend, backend, data, AI, agents, networking, and cybersecurity interviews. It diagnoses gaps one question at a time, retrieves evidence from connected MCP sources or user-provided material, and keeps probing until the learner can explain, apply, debug, compare, and transfer the concept without material hints.
The optional Obsidian integration stores durable learning state without moving execution into Markdown. Codex remains the reasoning and teaching layer; the Vault is a compact, navigable memory system.
- Mastery over coverage: completion requires observable evidence, not “I understand.”
- Progressive disclosure: one branch and one question at a time; hints escalate only when needed.
- Interview-oriented depth: mechanism, implementation, debugging, tradeoffs, failures, security, and novel transfer.
- Evidence-grounded learning: connected MCP sources first, then official documentation, standards, papers, repositories, files, or URLs.
- Job-description targeting: prioritize the concepts and project difficulties that matter to a specific role.
- Three learning tracks: technical interviews, resume/project defense, and behavioral interviews.
- Obsidian-ready continuity: resume across Codex tasks through compact topic, concept, session, source, and target notes.
- Duplicate-resistant concepts: canonical IDs, aliases, domains, and resolution checks keep the knowledge graph consistent.
- Safe by default: source repositories are read-only unless the user explicitly authorizes changes; cybersecurity work requires an authorized, isolated target.
flowchart LR
A["Topic, job description, repo, folder, or URL"] --> B["Codex + Super Study"]
B --> C["Evidence retrieval"]
C --> D["Diagnose one knowledge branch"]
D --> E["Explain or hint progressively"]
E --> F["Retest with a new scenario"]
F -->|"gap remains"| D
F -->|"evidence passes"| G["Mastery matrix"]
G --> H["Optional Obsidian checkpoint"]
H --> I["Resume from the exact unresolved branch"]
Super Study verifies recall, causal mechanism, practical use, boundaries, failure modes, tradeoffs, security implications, transfer, and interview follow-ups. A topic is complete only when every material dimension has evidence.
Give Codex this repository URL and ask:
Install the super-study skill from
https://github.com/Loffee5422/super-study/tree/main/skills/super-study
Clone the repository, then copy skills/super-study into your personal Codex skills directory.
macOS or Linux:
git clone https://github.com/Loffee5422/super-study.git
mkdir -p ~/.codex/skills
cp -R super-study/skills/super-study ~/.codex/skills/Windows PowerShell:
git clone https://github.com/Loffee5422/super-study.git
New-Item -ItemType Directory -Force "$env:USERPROFILE\.codex\skills" | Out-Null
Copy-Item -Recurse ".\super-study\skills\super-study" "$env:USERPROFILE\.codex\skills\"Restart Codex after installation so the Skill catalog refreshes.
Invoke the Skill explicitly with $super-study, or describe a matching learning goal naturally.
Use $super-study to help me master database indexing for a backend interview.
Use $super-study to analyze this job description and build a mastery path around its highest-value requirements: [paste JD]
Use $super-study to deeply defend the architecture and tradeoffs in this GitHub project: [repository URL]
Use $super-study to turn my experience into strong behavioral interview stories without inventing details.
You can provide a GitHub repository, local project folder, file, pasted link, job description, or resume project. If no source is supplied, the Skill prefers available MCP or connected search sources and authoritative primary material.
The Skill works without Obsidian. To preserve learning state across tasks, configure a Vault with the included deterministic CLI:
python skills/super-study/scripts/super_study.py configure \
--vault "/absolute/path/to/your/vault" --consent --create
python skills/super-study/scripts/super_study.py init
python skills/super-study/scripts/super_study.py validateThe generated Vault uses small navigation indexes and loads only the current Topic, latest relevant Session, and needed Concepts or Sources. It never requires importing the full Vault into context.
00 Home/ navigation and usage guide
10 Topics/ learning maps and mastery contracts
20 Concepts/ canonical reusable concepts
30 Sessions/ append-only learning evidence
40 Sources/ URLs, repositories, documents, and versions
50 Reviews/ review records and spaced-retrieval evidence
60 Targets/ job descriptions and interview targets
90 Templates/ reusable note templates
.super-study/ disposable machine indexes
The Vault path and consent state live in a local user configuration file. They are not included in this repository.
skills/super-study/
├── SKILL.md routing and core behavior
├── agents/openai.yaml Codex interface metadata
├── references/ lazily loaded learning protocols
├── scripts/super_study.py deterministic Vault operations
└── assets/vault-templates/ Obsidian notes and user guide
The modular design keeps the base Skill small. Codex reads only the track and reference material required for the current request, which reduces input-token use and makes the system easier to maintain.
Super Study prioritizes common concepts and realistic project difficulties that appear in software engineering work and interviews. It intentionally deprioritizes obscure trivia and excessively narrow details unless a job description, project, or source makes them relevant.
This project is a learning tool, not a guarantee of interview results. For cybersecurity topics, use only systems you own or are explicitly authorized to test.
Issues, learning-track proposals, protocol improvements, and reproducible bugs are welcome. Read CONTRIBUTING.md and SECURITY.md before contributing.
Released under the MIT License.