Prepare for a coding interview with your coding agent: turn a target role and interview requirements into a focused training plan, original practice tasks, a suitable local environment, and evidence-based feedback.
For candidates preparing with Codex, Claude Code, or another repository-aware coding agent. Covers industrial programming, algorithms, debugging and performance, ML/research engineering, design, and mixed interview formats. The role, language, seniority, platform, and AI policy come from your interview.
Use it if you want an agent to run preparation tailored to your upcoming interview, remember progress between sessions, and review more than whether tests pass. You can start with an incomplete job description and refine the plan together.
This is a small collection of agent instructions and working helpers. It is not a course, a question bank of real employer problems, an assessment simulator with validated scores, or a universal execution framework. Non-Python preparation uses the workflow; your agent supplies the tooling your case needs.
| Part | What it provides |
|---|---|
| Agent workflow | Interview intake, evidence tracking, training, mocks, feedback, and session memory |
| Interview specification | Editable Markdown for role, requirements, stages, timeboxes, tools, and AI policy |
| Python test runner | Standard-library tests, public/candidate/hidden groups, four cumulative levels, timeouts, and frozen runs |
| Optional VS Code environment | Public-test button, ordinary Python completion and lint, AI disabled, isolated editor data |
| Local practice history | First-edit timer, edit snapshots, exact tested versions, and review exports |
| Resolution example | Sanitized senior Research Engineer setup, practice ideas, and review lessons |
| Three original exercises | Artifact catalog, temporal settings, and job dispatcher: prompts, blank starters, and tests |
The exercises show one progressive industrial-coding format. They are optional examples, not the default curriculum for every role. Their private tests are published for authors and reviewers; candidates should avoid reading them during an unassisted mock. Worked candidate submissions and personal history are excluded.
Use this repository as a template, clone it, or fork it. Open the resulting workspace with your coding agent, and send:
Read AGENTS.md and help me prepare for my upcoming coding interview. First ask about the role, requirements, interview format, time available, permitted tools, and AI policy. Propose a training plan before assigning a mock. Use the bundled helpers where they fit, and build only what my preparation needs.
Bring the job description and any shareable interview instructions. The agent
maintains your setup and progress in kb/. The Resolution example
shows the kind of specification it can build.
The Markdown workflow requires no runtime. The optional runner needs Python 3.10+; VS Code setup needs VS Code 1.100+ and internet for the initial Pyright/Ruff installation. Recorder development tests use Node.js. The editor workflow has been verified on macOS; other platforms have not been verified.
Your starting points are AGENTS.md, the interview specification, and the helper reference. Beyond this README, the documentation is written for the agent managing your preparation.
New personal notes, new exercise attempts, reviews, editor data, and recordings are ignored by default. Bundled starters and memory bootstrap files are already tracked; review their later edits before publishing. Keep sensitive preparation local. If you publish an example, copy only deliberately sanitized material; source messages and confidential assessment content do not belong in a public fork. See publication guidance.
Original practice material is unaffiliated with any employer or assessment provider. Reusable code and documentation are available under the MIT license.