一键导入
repo-template
repo-template 收录了来自 irudik 的 40 个 skills，并提供仓库级职业覆盖和站内 skill 详情页。
这个仓库中的 skills
Recover context after compression or a new session. Reads MEMORY.md, recent plans, session logs, and git state to reconstruct working context.
Recover context after compression or a new session. Reads MEMORY.md, recent plans, session logs, and git state to reconstruct working context.
Save a durable, project-specific lesson to MEMORY.md in a structured [LEARN] format.
Trace an ambiguous result or failure back to its most likely cause using competing hypotheses and evidence. Produces a diagnostic report rather than a code change by default.
Save a durable, project-specific lesson to MEMORY.md in a structured [LEARN] format.
Trace an ambiguous result or failure back to its most likely cause using competing hypotheses and evidence. Produces a diagnostic report rather than a code change by default.
Stage, commit, create PR, and merge to main. Use for the standard commit-PR-merge cycle.
Compare script outputs between two git branches. Uses worktrees to run scripts on both branches and checksums to detect differences.
End-to-end R data analysis workflow from exploration through regression to publication-ready tables and figures
Audit MATLAB optimization derivatives and KNITRO integration with finite-difference validation
Run expert proofreading on academic documents. Checks grammar, typos, overflow risks, citation consistency, and academic quality for LaTeX, Quarto, and directly provided text-based files. Produces a report without editing files.
Verify-refactor-verify loop for safe code style changes. Records baseline checksums, applies style-only transformations, then verifies identical output.
Review and clean up comments, docstrings, and documentation across specified files or directories
Run the substantive domain review on manuscripts, slides, or lecture materials. Checks identification, derivations, citations, code-theory alignment, and logical consistency. Produces a report without editing files.
Run the Julia code review protocol on Julia scripts. Checks code quality, type stability, performance, and professional standards. Produces a report without editing files.
Run the Makefile review protocol. Checks conventions, dependency correctness, script coverage, and build system quality. Produces a report without editing files.
Run the MATLAB code review protocol on MATLAB scripts. Checks code quality, solver configuration, derivative correctness, and optimization patterns. Produces a report without editing files.
Review and address unresolved PR comments. Implements fixes, commits modularly, replies with commit hashes, and resolves threads.
Run the R code review protocol on R scripts. Checks code quality, reproducibility, domain correctness, and professional standards. Produces a report without editing files.
Run the Stata code review protocol on Stata scripts. Checks code quality, reproducibility, data-state safety, and professional standards. Produces a report without editing files.
Run the LaTeX review protocol on manuscript and slides. Detects hardcoded numeric results, distinguishes manuscript from slide contexts, and auto-fixes unambiguous values when the source is clear.
Generate a Makefile for a code directory by scanning scripts for output paths and building dependency rules.
Checksum script outputs before and after changes. Establishes baseline checksums, re-runs scripts, and compares to detect regressions. Uses CSV as the gold standard.
Stage, commit, create PR, and merge to main. Use for the standard commit-PR-merge cycle.
Compare script outputs between two git branches. Uses worktrees to run scripts on both branches and checksums to detect differences.
End-to-end R data analysis workflow from exploration through regression to publication-ready tables and figures
Audit MATLAB optimization derivatives and KNITRO integration with finite-difference validation
Run expert proofreading on academic documents. Checks grammar, typos, overflow risks, citation consistency, and academic quality for LaTeX, Quarto, and directly provided text-based files. Produces a report without editing files.
Verify-refactor-verify loop for safe code style changes. Records baseline checksums, applies style-only transformations, then verifies identical output.
Review and clean up comments, docstrings, and documentation across specified files or directories
Run the substantive domain review on manuscripts, slides, or lecture materials. Checks identification, derivations, citations, code-theory alignment, and logical consistency. Produces a report without editing files.
Run the Julia code review protocol on Julia scripts. Checks code quality, type stability, performance, and professional standards. Produces a report without editing files.
Run the Makefile review protocol. Checks conventions, dependency correctness, script coverage, and build system quality. Produces a report without editing files.
Run the MATLAB code review protocol on MATLAB scripts. Checks code quality, solver configuration, derivative correctness, and optimization patterns. Produces a report without editing files.
Review and address unresolved PR comments. Implements fixes, commits modularly, replies with commit hashes, and resolves threads.
Run the R code review protocol on R scripts. Checks code quality, reproducibility, domain correctness, and professional standards. Produces a report without editing files.
Run the Stata code review protocol on Stata scripts. Checks code quality, reproducibility, data-state safety, and professional standards. Produces a report without editing files.
Run the LaTeX review protocol on manuscript and slides. Detects hardcoded numeric results, distinguishes manuscript from slide contexts, and auto-fixes unambiguous values when the source is clear.
Generate a Makefile for a code directory by scanning scripts for output paths and building dependency rules.
Checksum script outputs before and after changes. Establishes baseline checksums, re-runs scripts, and compares to detect regressions. Uses CSV as the gold standard.