개발 스킬

Diagnose failing tests in the Morphia project by running the test, examining the full error, and tracing behavior through source and generated code. Make a minimal fix, then rerun the failing and related tests to check the result.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Finding the cause of a failing Morphia test
• Tracing an error into generated bytecode
• Checking whether an AOT or runtime generation path failed

Use this skill when technical design requirements seem to conflict. It helps name the contradiction, test whether time, space, conditions, or scale can separate the opposing states, and propose a concrete design that preserves both benefits where possible.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Resolving competing system architecture requirements
• Finding a design that balances freshness and caching
• Separating strictness from ease of use

When documentation, tests, or assumptions conflict with observed behavior, this skill directs you to inspect current code or data, record the concrete difference, and decide whether the implementation, the description, or both need updating.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Checking a mismatch between docs and behavior
• Investigating a passing test with a production issue
• Verifying whether a metric or schema is current

Use this skill to compare long-lived ideas, technologies, and practices when durability matters. It treats survival age as a clue to future longevity, while checking whether the field has changed and whether a newer option solves a necessary gap the proven choice cannot address.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Comparing frameworks for a long-lived system
• Choosing a protocol or practice for lasting use
• Checking whether a newer option solves a real gap

Use this skill when a proposed fix adds a feature, dependency, or process. It helps identify evidence-backed removals first, protect important safeguards, and check the result before deciding whether any small addition is still needed.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Simplifying a complex workflow
• Reviewing a proposed feature addition
• Removing a demonstrably unused dependency

Assess the impact of a change to a shared interface, setting, or data format by finding every consumer and running representative end-to-end paths. Compare prior outputs to catch broken assumptions, unexpected result changes, and downstream work that needs rebuilding or repinning.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Change a shared function signature or return value safely
• Add or rename a data column and check downstream consumers
• Verify a pipeline change that affects reported numbers

Checks the current work against quality and consistency gates, stages specific files, and creates a descriptive commit on a branch. It can push and open a pull request when requested, but it does not merge changes.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Committing finished changes after quality checks
• Creating a branch before committing work
• Opening a pull request with the commit when requested

Diagnose why Claude Code asks for tool approval by checking permission settings and their precedence across repository and session layers. The read-only workflow starts with project-local files; it accesses host-wide settings only after explicit user confirmation.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Finding why Claude Code keeps asking for approval
• Checking why a permission bypass setting has no effect
• Comparing project settings with the active session mode

Use this skill to check review comments, audit findings, or critiques against the actual source. It classifies each point as confirmed, refuted, or overstated, supports decisions with evidence, and helps apply verified fixes without introducing errors.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Checking an AI reviewer’s comments against the source
• Sorting audit or code review findings by severity
• Applying only fixes supported by evidence

Review `.R` scripts for code quality, reproducibility, domain correctness, tidyverse conventions, and professional standards. It produces a findings report while leaving source files unchanged; it does not run the code or verify numerical results.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Reviewing an R script before submission
• Checking analysis code quality and reproducibility
• Finding issues in tidyverse usage

Create a Git commit from current changes using the session history to explain their purpose. The skill checks which files belong, stages the intended work, and writes a structured message covering changes, rationale, and validation.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Commit completed work with its rationale
• Draft a commit message from the session history
• Check staged files for unrelated changes

Update the current feature branch with the latest `origin/main` using a merge, then guide conflict resolution by comparing each side’s intent. It includes steps for checking repository state, handling conflicts carefully, and reviewing the completed merge.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Sync a feature branch with the latest main
• Resolve conflicts during a merge-based update
• Review a branch update before continuing work

Debug Borg behavior inside a running Docker container by executing real Borg 1 and Borg 2 commands and inspecting their output. Use the results to write accurate code, tests, and error handling based on actual command responses.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Checking real Borg command output before coding
• Testing against Borg 1 or Borg 2 behavior
• Creating a temporary repository to reproduce an issue

Continue an engineering specification from its progress table by carrying out the next planning, implementation, or review step in the current session. It follows the spec’s named model and pauses at its decision gates.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Continuing the next phase of a multi-step engineering spec
• Finding out which spec phase should happen next
• Running the spec’s planning, implementation, or review step

Guide a pull request through conflict checks, reviewer feedback, and CI results, fixing issues and waiting for required checks to finish. When the PR is ready, squash-merge it using its title and description.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Landing a pull request after CI checks pass
• Resolving merge conflicts before a PR is merged
• Monitoring review feedback and checks through completion

Push the current branch to its configured remote and create a pull request if one is missing, or update the open request to match the branch’s full scope. It checks changes and handles sync problems through the related pull workflow.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Publishing the current branch and opening a pull request
• Updating a pull request after adding branch changes
• Pushing again after resolving a branch sync issue

Review a GitHub pull request through several focused code analysis angles, then verify and rank candidate findings. The workflow filters for well-supported, significant issues and produces actionable review feedback; it is intended for GitHub PRs, not local uncommitted changes.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Reviewing a GitHub pull request before merging
• Checking a PR for concrete bugs and security risks
• Rechecking a PR after new commits arrive

Handle a GitHub issue through a structured workflow: inspect its details, research related code and pull requests, plan and implement a fix, run tests, then submit a pull request that references the issue.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Fix a bug reported in a GitHub issue
• Investigate an issue and plan the changes
• Implement and test an issue fix

Use this skill to shape a feature, component, or system before implementation. It asks focused questions, compares design options, and helps produce an agreed specification so the work starts with clearer requirements and decisions.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Clarifying requirements for a new feature
• Comparing design options before implementation
• Breaking a large project into smaller design efforts

Create or revise agent skills, then test how they perform on realistic prompts. Review outputs and benchmark results, use feedback to refine the skill, and improve its description so agents can recognize when to use it.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Draft a skill for a repeated workflow
• Evaluate an existing skill with test prompts
• Improve a skill based on user feedback

Use this skill for a focused, challenging interview about a proposal or design. It asks one question at a time, tests assumptions against concrete scenarios and code, and records agreed terms and significant decisions in project context files and ADRs.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Stress-testing assumptions in a design proposal
• Clarifying ambiguous domain terms
• Checking a design against existing code

Delegate coding, code review, and implementation plan review to the OpenAI Codex CLI. Choose a sandbox level for the task, then relay Codex’s findings or implementation results; review findings are presented for the user to decide on.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Delegate a coding task to Codex
• Request a second opinion on code
• Review an implementation plan

Review a completed feature, work session, or incident using concrete evidence such as failed checks and rework. Identify costly friction, trace likely causes, and propose durable fixes, prioritizing automation and keeping only a few high-impact findings.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Reviewing a completed feature or development session
• Finding lessons from an incident or failed check
• Turning repeated rework into an automation or process fix

Plan and guide structural refactors across files or modules while preserving behavior and public contracts. The skill maps dependencies and consumers, ranks changes by priority, and lays out staged moves with validation steps and compatibility considerations.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Reorganizing modules across multiple files
• Planning a staged package or boundary refactor
• Reviewing API and compatibility risks