master-learning is a Codex skill for pre-project domain learning. It helps Codex research unfamiliar domains before planning or implementing, especially when a task depends on current documentation, papers, standards, GitHub repositories, examples, issues, tests, or local code conventions.

The core idea is simple: a master keeps the mind of an apprentice. Before building, learn the field.

AI coding agents are fast, but unfamiliar domains often punish speed. If an agent starts coding before reading current docs, paper assumptions, GitHub examples, local project conventions, and known failure modes, the result can be technically plausible but wrong.

master-learning adds a repeatable research gate before implementation. It turns vague uncertainty into a source-backed Learning Brief, so Codex can plan and build from evidence instead of stale memory or guesses.

Many coding agents can write code quickly, but they often fail when the task requires learning first. A new framework, a paper-backed method, a GitHub ecosystem, a changing API, or a project with strong local conventions can make direct implementation risky.

master-learning adds a disciplined "learn before build" phase to Codex:

• Inspect local project files, dependency manifests, configs, tests, and existing conventions.
• Study official documentation, release notes, migration notes, standards, and specifications.
• Read papers and translate methods, assumptions, and evaluation setup into engineering constraints.
• Review GitHub repositories beyond stars: license, activity, examples, tests, issues, dependency health, and reuse risk.
• Mark weak evidence, stale APIs, source conflicts, abandoned repositories, missing licenses, and provisional conclusions.
• Produce a decision-ready Learning Brief before implementation begins.

The final Learning Brief becomes the handoff artifact for implementation. It explains what to build, what to avoid, which sources support the plan, what risks remain, and what acceptance criteria should be used.

Use $master-learning when you are:

• Starting a project in an unfamiliar technical or research domain.
• Choosing a framework, library, algorithm, paper method, or architecture.
• Asking Codex to follow latest docs, best practices, standards, or GitHub examples.
• Adapting a GitHub project into your own project.
• Working on a task where a wrong assumption would waste implementation time.
• Modifying a local project where existing conventions should be read before editing.

Do not use it for trivial edits, typo fixes, formatting-only work, or direct bug fixes with clear local evidence.

The brief is designed for real engineering work, not for decorative research:

• Task: user goal, target environment, success criteria, research depth, and confidence.
• Sources: URLs or paths, source type, date or currency, reliability, and purpose.
• Domain Model: key concepts, objects, data, relationships, and vocabulary.
• Local Code Lessons: repository structure, conventions, configs, tests, and constraints.
• GitHub/Code Lessons: reviewed repositories, patterns, license and reuse notes, examples, and issues.
• Paper/Standard Lessons: methods, assumptions, evaluation setup, requirements, and limits.
• Implementation Patterns: architecture, API contracts, data flow, control flow, and test strategy.
• Risks and Anti-Patterns: weak assumptions, edge cases, stale docs, source conflicts, and things to avoid.
• Recommendation: proposed approach, acceptance criteria, and next steps.
• Open Questions: unresolved decisions or missing evidence.

This repository includes a Microsoft SkillOpt-inspired local optimization workflow. It does not fine-tune a model. Instead, it treats SKILL.md as the trainable artifact, applies bounded text edits, and accepts a candidate only after validation.

Included artifacts:

• master-learning/references/skillopt-training.md
• master-learning/scripts/skillopt_train.py
• master-learning/training/benchmark-scenarios.json
• master-learning/training/skillopt-run-2026-06-21.md
• master-learning/training/skillopt-run-2026-06-21-round2.md
• master-learning/training/skillopt-run-2026-06-21-128.md

Scenario coverage:

• Latest framework and API tasks
• Paper reproduction tasks
• GitHub adaptation tasks
• Local-project-first tasks
• Low-risk skip behavior
• Network-degraded research

The 128-iteration stability run reached score 1.0 and passed the release gate.

Clone this repository, then copy the master-learning skill folder into your Codex skills directory.

git clone https://github.com/GzmSTAR/master-learning-skill.git
Copy-Item -Recurse -Force .\master-learning-skill\master-learning "$env:USERPROFILE\.codex\skills\master-learning"

git clone https://github.com/GzmSTAR/master-learning-skill.git
mkdir -p ~/.codex/skills
cp -R master-learning-skill/master-learning ~/.codex/skills/master-learning

Restart Codex if the skill list does not refresh automatically.

ls ~/.codex/skills/master-learning

Use $master-learning to study robot vision SLAM libraries, produce a Learning Brief, then plan the implementation.

Use $master-learning before building this paper reproduction project. Check official docs, papers, GitHub repos, and known failure modes.

master-learning/
  SKILL.md
  agents/openai.yaml
  references/
  scripts/
  training/

The helper scripts use only the Python standard library:

• create_learning_brief.py
• github_scan.py
• paper_scan.py
• source_audit.py
• merge_learning_brief.py
• skillopt_train.py

python "$env:USERPROFILE\.codex\skills\.system\skill-creator\scripts\quick_validate.py" "$env:USERPROFILE\.codex\skills\master-learning"
python "$env:USERPROFILE\.codex\skills\master-learning\scripts\skillopt_train.py" --help

Claude Code version: https://github.com/GzmSTAR/master-learning-claude-code-skill

MIT