Skills

Reusable agent skills for project setup and development workflows.

Contributing

See CONTRIBUTING.md for contribution expectations and pull request verification.

NPX Installer

This repository ships a dependency-free Node CLI for installing every skill into agent CLIs and IDE rule folders.

Run the guided installer:

npx @jeremymcs/skills

Install every skill for Codex:

npx @jeremymcs/skills install --target codex --all

Install selected skills into project-local agent skills:

npx @jeremymcs/skills install --target agents --skill dumb-it-down,zero-techdebt

Export a skill as Cursor rules with a model profile hint:

npx @jeremymcs/skills install --target cursor --skill github-repo-hygiene --model gpt-5

Installer Targets

| Target | Output | | --- | --- | | codex | Native skill folders in ~/.codex/skills or $CODEX_HOME/skills | | claude | Native skill folders in ~/.claude/skills | | agents | Native project skill folders in ./.agents/skills | | cli | Portable native skill folders in ./.agent-skills | | cursor | Cursor rule files in ./.cursor/rules/skills plus bundled skill assets | | windsurf | Windsurf rule files in ./.windsurf/rules/skills plus bundled skill assets | | continue | Continue rule files in ./.continue/rules/skills plus bundled skill assets | | all | Installs to every target above |

The native targets copy each skill directory exactly, including SKILL.md, metadata, scripts, references, and templates. IDE rule targets create one rule file per skill and keep the full skill source beside the rules in _skills/<skill>/.

Installer Options

npx @jeremymcs/skills list
npx @jeremymcs/skills targets
npx @jeremymcs/skills install --target codex,cursor --all --force
npx @jeremymcs/skills install --target cli --dest /tmp/agent-skills --dry-run

Supported model profiles are auto, gpt-5, gpt-4.1, claude-sonnet, claude-opus, gemini-pro, and local. Model profiles are hints for hosts that surface model-specific instructions; native skill installs preserve the original skill content.

Available Skills

| Skill | Purpose | Use when | | --- | --- | --- | | $create-issue | Creates a single post-mortem issue and pull request from completed session work. | Work is already done and needs to be captured in the repo issue tracker, committed, pushed, and opened as a PR. | | $dumb-it-down | Turns confusing ideas into plain-language explanations with simple references. | You want dense writing, jargon, code, contracts, technical topics, or abstract ideas broken down so they are extremely easy to understand. | | $falsify-project | Attempts to disprove project claims before validating the repo, app, or site. | You want an adversarial audit that treats docs and claims as false until code/runtime evidence survives contradiction checks. | | $from-issue | Researches an existing issue, asks only codebase-unanswerable questions, implements a patch, verifies it, and opens a PR. | You want Codex to take a GitHub issue, tracker ticket, bug report, feature issue, or local issue artifact from investigation through pull request. | | $github-repo-hygiene | Audits GitHub repository, package, and dependency hygiene. | You want evidence-backed recommendations for package manifests, dependencies, workspaces, docs, CI, security, contribution flow, branch settings, releases, and repo cleanliness. | | $new-feature | Researches the current project and recommends prioritized new or next features. | You want evidence-backed feature ideas from the current repo, docs, issues, and recent work. | | $overengineering-review | Reviews code for avoidable complexity and writes an HTML findings report. | You want code inspected for too many layers, premature abstraction, generic machinery, unclear indirection, or complexity beyond the current requirement. | | $project-doc-scaffolder | Scaffolds agent-ready project documentation for a new or existing repository. | A project needs AGENTS.md, contributor docs, design notes, issue templates, and a PR template. | | $system-diagrammer | Walks code slowly to diagram and verify system, data, workflow, and integration connections. | You need a careful code-backed map of how systems connect, including findings, evidence, recommended fixes, breaks, edge cases, and disconnects. | | $test-writer | Enforces strict hard tests for every non-documentation change. | You want each change proven by executable behavior tests, RED/GREEN output, and sabotage failure rather than grep, source-text checks, snapshots-only assertions, or manual verification. | | $to-adr | Turns an architecture discussion or design decision into a complete ADR. | A conversation has clarified a decision, rejected option, tradeoff, or architecture direction that future contributors should not re-litigate. | | $verify-project | Verifies whether a project, repo, or app actually does what it claims. | You need a code-backed completion report with evidence, gaps, and next steps. | | $zero-techdebt | Reviews code changes for function scope, naming, and file organization fit. | You want to know whether a refactor or code change fits the existing codebase without adding shallow abstractions or confusing structure. |

Project Doc Scaffolder

$project-doc-scaffolder creates:

• AGENTS.md
• CONTRIBUTING.md
• DESIGN.md
• .github/ISSUE_TEMPLATE/bug_report.md
• .github/ISSUE_TEMPLATE/feature_request.md
• .github/pull_request_template.md

Usage

Run the project doc scaffold script from this repository:

python3 project-doc-scaffolder/scripts/scaffold_project_docs.py /path/to/project

Existing files are skipped by default. To overwrite existing files after reviewing the impact:

python3 project-doc-scaffolder/scripts/scaffold_project_docs.py /path/to/project --force

Skill Layout

Each skill lives in its own folder and includes a required SKILL.md. Supporting templates, scripts, and metadata live under that skill folder.

Current layout:

create-issue/
├── SKILL.md
└── agents/openai.yaml

dumb-it-down/
├── SKILL.md
├── agents/openai.yaml
└── references/simplification-patterns.md

falsify-project/
├── SKILL.md
└── agents/openai.yaml

github-repo-hygiene/
├── SKILL.md
├── agents/openai.yaml
└── scripts/audit_package_hygiene.mjs

from-issue/
├── SKILL.md
└── agents/openai.yaml

new-feature/
├── SKILL.md
└── agents/openai.yaml

overengineering-review/
├── SKILL.md
└── agents/openai.yaml

system-diagrammer/
├── SKILL.md
└── agents/openai.yaml

test-writer/
├── SKILL.md
└── agents/openai.yaml

to-adr/
├── SKILL.md
└── agents/openai.yaml

verify-project/
├── SKILL.md
└── agents/openai.yaml

zero-techdebt/
├── SKILL.md
└── agents/openai.yaml

project-doc-scaffolder/
├── SKILL.md
├── agents/openai.yaml
├── assets/templates/
└── scripts/scaffold_project_docs.py