jungho-git
82 verified skills82 total stars
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
tools1
skill-installer
Install Claude skills into `~/.claude/skills` or a custom directory. Use when a user asks to list installable skills, install a skill from the default `anthropics/skills` repository, or install a skill from another GitHub repo or path (including private repos).
tools1
openai-docs
Use when the user asks how to build with OpenAI products or APIs and needs up-to-date official documentation with citations, help choosing the latest model for a use case, or explicit GPT-5.4 upgrade and prompt-upgrade guidance; prioritize OpenAI docs MCP tools, use bundled references only as helper context, and restrict any fallback browsing to official OpenAI domains.
tools1
skill-installer
Install Codex skills into $CODEX_HOME/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).
development1
execution
Smallest complete change rule: preserve local code shape, extend existing patterns, avoid speculative extraction or cleanup, and include required coupled updates for correctness.
development1
markdown
Markdown rules for documentation and skill files: ATX headings, consistent hierarchy, fenced code blocks with language tags, H2 dividers in maintained instruction docs, and predictable rendering.
development1
benchmark
Agent benchmark workflow for comparing token usage, latency, cost, cache behavior, model choice, prompt variants, tool availability, and configuration changes with fixed tasks and measured evidence.
tools1
hooks
Hook infrastructure alignment: treat configured hooks as required workspace infrastructure, validate via real entrypoint.
testing1
ignore
Hard read/edit boundary for generated, cached, temporary, and non-source-of-truth paths.
tools1
xlsx
Spreadsheet workflow for reading, editing, fixing, creating, validating, and delivering .xlsx/.xlsm/.csv/.tsv files, with template preservation, formula-first calculations, recalc checks, formatting rules, and zero formula errors.
development1
fs-mcp-failure-recovery
Use when the user invokes $fs-mcp-failure-recovery with a YYYY-MM-DD date to analyze that date %USERPROFILE%/.mcp/fs-mcp-failures log, identify unresolved failure groups, modify the real affected projects, verify fixes, and move verified items into a separate fs-mcp success JSONL list.
tools1
comments
Code comment policy: numbered one-line `―` dividers for touched declarations and logical sections, paired outer blocks only for long regions, concise purpose comments, and no comment churn.
development1
memory
Search and recall relevant prior-session memories, earlier decisions, past debugging context, rollout notes, and durable project knowledge. Use only for questions about prior sessions or historical context, not current-code inspection.
development1
failure
Failure analysis for runtime errors, failing tests, broken builds, regressions, and incorrect behavior: reproduce the failing path, capture the first symptom, isolate the first break point, and patch only after evidence supports the cause.
development1
boundary
Source-of-truth boundary for generated, cached, temporary, vendor, build, log, and non-authoritative paths. Use when deciding what local files are valid inputs for exploration, edits, review, or validation, and how to handle explicit exceptions.
development1
surface
Surface analysis before implementation or review: map entry points, callers, dependencies, contracts, schemas, data flow, configs, generated artifacts, and cross-project edges without drifting into whole-repo exploration.
testing1
rag
Plan, scaffold, document, index, evaluate, or extend a retrieval-augmented knowledge base for company knowledge, including project docs, table docs, process docs, glossary entries, metadata, chunking, retrieval checks, and rollout gates.
testing1
workflow
Required phase order for non-trivial tasks: Plan, Explore, Implement, Verify, Finalize. Use for multi-step work, scoped exploration, re-planning, validation, and final synthesis.
testing1
sql
SQL and MyBatis rules for queries, mapper XML, dynamic SQL, and database contract edges: minimize joins, preserve result semantics, keep business logic in Java, and validate parameters, aliases, plans, and edge data.
development1
yaml
YAML rules for application config, CI/CD, Kubernetes, Docker Compose, and skill agent manifests: 2-space indent, quoted ambiguous strings, block scalars, no plain secrets, and schema-aware validation.
testing1
json
JSON rules for package manifests, tsconfig, API schemas, config files, and payloads: double-quoted keys, no comments, no trailing commas, stable order, 2-space indent, and schema-aware validation.
development1
xml
XML and MyBatis mapper rules for namespaces, schema references, mapper contracts, embedded SQL, dynamic branches, and well-formedness validation.
data-ai1
benchmark
Agent benchmark workflow for comparing token usage, latency, cost, cache behavior, model choice, prompt variants, tool availability, and configuration changes with fixed tasks and measured evidence.
tools1
review
Read-only review: correctness bugs first, then regression risks, commit issues, expected failure points, missing coupled files, and practical fix guidance.
data-ai1
markdown
Markdown rules for documentation and skill files: ATX headings, consistent hierarchy, fenced code blocks with language tags, H2 dividers in maintained instruction docs, and predictable rendering.
development1
validate
Post-implementation validation: scoped to changed surface, original failing path, coupled contracts, data/schema effects, and real command evidence over assumption-only claims.
testing1
memory-recall
Search and recall relevant prior-session memories, earlier decisions, past debugging context, rollout notes, and durable project knowledge. Use only for questions about prior sessions or historical context, not current-code inspection.
development1
hooks
Hook infrastructure alignment: treat configured hooks as required workspace infrastructure, validate via real entrypoint.
testing1
shell
Shell/Bash rules for scripts, pipelines, paths, and exit codes: quote variables, set -euo pipefail, no eval, arrays for arguments, explicit interpreter, and safe block formatting.
development1
codex-work-report
Generate or update a Korean Markdown report from Codex work results, session logs, completion-hook summaries, command checks, changed files, blockers, and follow-up actions. Use when the user asks for a Codex 작업 결과 보고서, 작업 로그 보고서, md 보고서, 결과 정리 문서, or to revise/version a previous Codex work report.
development1
validate
Post-implementation validation: scoped to changed surface, original failing path, coupled contracts, data/schema effects, and real command evidence over assumption-only claims.
testing1
response
Final response format: Korean-first, concise Process / Checks / Issues / Updates, optional Usage, with only actual changes, actual validation, real blockers, changed files, and measured token data when available.
development1
mybatis
MyBatis mapper rules for XML mapper files, mapper interfaces, statement contracts, parameter/result shapes, embedded SQL, dynamic branches, and binding validation.
data-ai1
regex
Regex rules for explaining, debugging, testing, converting, and improving regular expressions across flavors. Use when working with regex patterns, pattern matching, test strings, validation rules, or flavor-specific regex behavior.
development1
docs
Generate or update a Korean Markdown report from Codex work results, session logs, completion summaries, command checks, changed files, blockers, and follow-up actions, with evidence labels and versioned report handling.
development1
boundary
Source-of-truth boundary for generated, cached, temporary, vendor, build, log, and non-authoritative paths. Use when deciding what local files are valid inputs for exploration, edits, review, or validation, and how to handle explicit exceptions.
development1
rag
Plan, scaffold, document, index, evaluate, or extend a retrieval-augmented knowledge base for company knowledge, including project docs, table docs, process docs, glossary entries, metadata, chunking, retrieval checks, and rollout gates.
testing1
memory-recall
Search and recall relevant prior-session memories, earlier decisions, past debugging context, rollout notes, and durable project knowledge. Use only for questions about prior sessions or historical context, not current-code inspection.
development1
failure
Failure analysis for runtime errors, failing tests, broken builds, regressions, and incorrect behavior: reproduce the failing path, capture the first symptom, isolate the first break point, and patch only after evidence supports the cause.
development1
surface
Surface analysis before implementation or review: map entry points, callers, dependencies, contracts, schemas, data flow, configs, generated artifacts, and cross-project edges without drifting into whole-repo exploration.
testing1
response
Final response format: Korean-first, concise four-section Process / Checks / Issues / Updates, with only actual changes, actual validation, real blockers, and changed files.
development1
workflow
Required phase order for non-trivial tasks: Plan, Explore, Implement, Verify, Finalize. Use for multi-step work, scoped exploration, re-planning, validation, and final synthesis.
testing1
javascript
JavaScript rules for browser, frontend, Node/Bun scripts, and mixed JS/TS surfaces: const/let only, template literals, typed boundaries where present, vanilla JS preference, Bun runtime, cohesive functions, and real boundary validation.
development1
json
JSON rules for package manifests, tsconfig, API schemas, config files, and payloads: double-quoted keys, no comments, no trailing commas, stable order, 2-space indent, and schema-aware validation.
development1
sql
SQL and MyBatis rules for queries, mapper XML, dynamic SQL, and database contract edges: minimize joins, preserve result semantics, keep business logic in Java, and validate parameters, aliases, plans, and edge data.
development1
shell
Shell/Bash rules for scripts, pipelines, paths, and exit codes: quote variables, set -euo pipefail, no eval, arrays for arguments, explicit interpreter, and safe block formatting.
development1
xml
XML and MyBatis mapper rules for namespaces, schema references, mapper contracts, embedded SQL, dynamic branches, and well-formedness validation.
data-ai1
yaml
YAML rules for application config, CI/CD, Kubernetes, Docker Compose, and skill agent manifests: 2-space indent, quoted ambiguous strings, block scalars, no plain secrets, and schema-aware validation.
testing1
review
Read-only review: correctness bugs first, then regression risks, commit issues, expected failure points, missing coupled files, and practical fix guidance.
data-ai1
python
Python rules for scripts, services, CLIs, and typed data boundaries: type hints, pathlib, f-strings, dataclasses or TypedDict, specific exceptions, resource safety, and no index-based loops.
tools1
java
Java rules for services, controllers, repositories, DTOs, mapper interfaces, and Java/MyBatis contract edges: Java 11 baseline, explicit types, null safety at boundaries, cohesive methods, Stroustrup layout, and no @SuppressWarnings.
development1
docs
Generate or update a Korean Markdown report from Codex work results, session logs, completion summaries, command checks, changed files, blockers, and follow-up actions, with evidence labels and versioned report handling.
development1
comments
Code comment policy: numbered one-line `―` dividers for touched declarations and logical sections, paired outer blocks only for long regions, concise purpose comments, and no comment churn.
development1
notion
Create or update Korean Notion-ready work reports from git/work evidence using a numbered report format, meaningful file filtering, business-day periods, arrow-prefixed file sections, and page-safe Markdown.
development1
typescript
TypeScript rules for .ts/.tsx/.mts/.cts files: preserve typed boundaries, avoid any and broad assertions, validate external data, keep Bun/runtime compatibility, and preserve import/module shape.
development1
notion
Create or update Korean Notion-ready work reports from git/work evidence using a numbered report format, meaningful file filtering, business-day periods, arrow-prefixed file sections, and page-safe Markdown.
development1
pwsh
PowerShell rules for .ps1 scripts and native command orchestration: paths as data, fs-mcp-first file work, no Invoke-Expression, structured arguments, exit-code checks, and Stroustrup-style block layout.
tools1
public/codex/memories/skills/mcp-startup-smoke
--- name: mcp-startup-smoke description: Diagnose local MCP startup failures or install-and-test requests like "MCP failed to start", "connection closed: initialize response", or "playwright mcp 설치 및 테스트" using config, logs, resolved command shape, and a direct smoke. disable-model-invocation: true user-invocable: false allowed-tools: - Read - Grep - Bash --- # MCP Startup Smoke ## When to use - Use for local Codex MCP startup failures, wrapper-path drift, service/auth suspicions, or in
tools1
graphify
any input (code, docs, papers, images) → knowledge graph → clustered communities → HTML + JSON + audit report
development1
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Codex's capabilities with specialized knowledge, workflows, or tool integrations.
tools1
ignore
Hard read/edit boundary for generated, cached, temporary, and non-source-of-truth paths.
tools1
fs-mcp
Local workspace routing: use fs-mcp first for file operations, structured search, process sessions, config inspection, and explicit fs-mcp git workflows.
tools1
surface
Surface analysis before implementation or review: map entry points, callers, dependencies, contracts, schemas, data flow, configs, generated artifacts, and cross-project edges without drifting into whole-repo exploration.
testing1
python
Python rules for scripts, services, CLIs, and typed data boundaries: type hints, pathlib, f-strings, dataclasses or TypedDict, specific exceptions, resource safety, and no index-based loops.
tools1
typescript
TypeScript rules for .ts/.tsx/.mts/.cts files: preserve typed boundaries, avoid any and broad assertions, validate external data, keep Bun/runtime compatibility, and preserve import/module shape.
development1
failure
Failure analysis for runtime errors, failing tests, broken builds, regressions, and incorrect behavior: reproduce the failing path, capture the first symptom, isolate the first break point, and patch only after evidence supports the cause.
development1
rag
Plan, scaffold, document, index, evaluate, or extend a retrieval-augmented knowledge base for company knowledge, including project docs, table docs, process docs, glossary entries, metadata, chunking, retrieval checks, and rollout gates.
testing1
execution
Smallest complete change rule: preserve local code shape, extend existing patterns, avoid speculative extraction or cleanup, and include required coupled updates for correctness.
development1
benchmark
Agent benchmark workflow for comparing token usage, latency, cost, cache behavior, model choice, prompt variants, tool availability, and configuration changes with fixed tasks and measured evidence.
tools1
comments
Code comment policy: numbered one-line `―` dividers for touched declarations and logical sections, paired outer blocks only for long regions, concise purpose comments, and no comment churn.
development1
boundary
Source-of-truth boundary for generated, cached, temporary, vendor, build, log, and non-authoritative paths. Use when deciding what local files are valid inputs for exploration, edits, review, or validation, and how to handle explicit exceptions.
development1
comments
Code comment policy: numbered one-line `―` dividers for touched declarations and logical sections, paired outer blocks only for long regions, concise purpose comments, and no comment churn.
development1
boundary
Source-of-truth boundary for generated, cached, temporary, vendor, build, log, and non-authoritative paths. Use when deciding what local files are valid inputs for exploration, edits, review, or validation, and how to handle explicit exceptions.
development1
workflow
Required phase order for non-trivial tasks: Plan, Explore, Implement, Verify, Finalize. Use for multi-step work, scoped exploration, re-planning, validation, and final synthesis.
testing1
response
Final response format: Korean-first, concise Process / Checks / Issues / Updates, optional Usage, with only actual changes, actual validation, real blockers, changed files, and measured token data when available.
development1
xlsx
Spreadsheet workflow for reading, editing, fixing, creating, validating, and delivering .xlsx/.xlsm/.csv/.tsv files, with template preservation, formula-first calculations, recalc checks, formatting rules, and zero formula errors.
development1
pwsh
PowerShell rules for .ps1 scripts and native command orchestration: paths as data, fs-mcp-first file work, no Invoke-Expression, structured arguments, exit-code checks, and Stroustrup-style block layout.
tools1
javascript
JavaScript rules for browser, frontend, Node/Bun scripts, and mixed JS/TS surfaces: const/let only, template literals, typed boundaries where present, vanilla JS preference, Bun runtime, cohesive functions, and real boundary validation.
development1
execution
Smallest complete change rule: preserve local code shape, extend existing patterns, avoid speculative extraction or cleanup, and include required coupled updates for correctness.
development1
java
Java rules for services, controllers, repositories, DTOs, mapper interfaces, and Java/MyBatis contract edges: Java 11 baseline, explicit types, null safety at boundaries, cohesive methods, Stroustrup layout, and no @SuppressWarnings.
development1
foo-skill
foo foo foo. Use when iospe bar baz.
tools1
fs-mcp-failure-recovery
Use when the user invokes $fs-mcp-failure-recovery with a YYYY-MM-DD date to analyze that date %dev-llm%/.mcp/fs-mcp-failures log, identify unresolved failure groups, modify the real affected projects, verify fixes, and move verified items into a separate fs-mcp success JSONL list.
tools1