the-skills

A small repo of AI agent skills — modular, each skill is self-contained with its own SKILL.md and references/.

License: MIT Repo: https://github.com/pxuanbach/the-skills

---

Skills

Skill	Description
init-agents	Analyzes repository structure, configurations, and conventions to generate a comprehensive AGENTS.md guide for AI agent collaboration.
read-paper	Read and analyze a single research paper using the 4-pass reading strategy and PACES framework. Cross-references with MEMORY.md and saves structured Markdown to docs/paper_summary/.
write-paper	Plan, draft, revise, and polish a scientific paper with a story-first method (Goal → Problem → Solution, the nugget), section guides, line-level style lint, a mock reviewer pass, and a LaTeX/submission checklist. Never invents results or citations.
research-workflow	Structured 4-phase research workflow (problem framing → info gathering → analysis → report drafting) with parallel data collection, citation-grounded outputs, and reusable reference data for software comparisons, paper deep-dives, and PDF batch extraction.
screenshot	Capture screenshots of web pages with stealth mode to bypass anti-bot detection. Saves images to .temp/ in the current workspace. Supports Cloudflare-protected sites.
review-skill	Review and validate Agent Skills for structural consistency, metadata accuracy, instruction clarity, and adherence to repo conventions.

New SDLC Workflow - RAD (6 skills)

A modular software development lifecycle workflow — each skill is self-contained and can be used independently or as a pipeline.

Skill	Role
requirement-analyzer	Gathers, clarifies, and formalizes software requirements and user stories into the LLM Wiki.
user-designer	Transforms approved requirements into technical design, UI/UX mockups and implementation plans (plan.md).
wiki-manager	Central documentation hub — manages, organizes, and queries all SDLC artifacts (requirements, designs, mockups, plans, evidence). Used by all other SDLC skills.
constructor	Executes implementation plans step-by-step — writes code, runs tests, and produces testing evidence in the LLM Wiki.
quality-reviewer	Reviews source code, implementation plans, and test evidence for quality, design integrity, and test coverage.
security-reviewer	Runs built-in Python scanners (FastAPI, Next.js, secrets), triages findings, audits 10 categories, issues pass/fail.

---

Repo convention

Each skill lives under skills/<skill-name>/:

skills/
└── <skill-name>/
    ├── SKILL.md          # Required: YAML frontmatter + workflow body
    ├── references/       # Optional: pre-researched data and reusable patterns
    │   └── *.md
    └── examples/         # Optional: example outputs
        └── *.md

When adding a new skill:

1. Create skills/<skill-name>/SKILL.md with valid YAML frontmatter (name, description, version, license).
2. Add references/ for context that the agent should load on demand.
3. Add a row to the table above (or to the SDLC Workflow table if part of that group).
4. Bump CHANGELOG.md under a new version section.

SDLC Workflow skills are designed to work together as a pipeline: requirement-analyzer → user-designer → constructor → quality-reviewer / security-reviewer, all coordinated through wiki-manager. When adding a skill to this group, ensure frontmatter name and description are precise — agents use these fields to select the right skill automatically.

---

Installation

Run a single command to launch an interactive selection menu. You can selectively choose which Coding Agents and which Skills to download and install (no git clone required):

Windows (PowerShell)

irm https://raw.githubusercontent.com/pxuanbach/the-skills/main/install.ps1 | iex

Non-interactive / CLI flags (optional):

# Install specific skills to specific agents
.\install.ps1 -Targets claude,antigravity -Skills wiki-manager,constructor

# Install all skills to all agents
.\install.ps1 -Targets all -Skills all

macOS / Linux (Bash)

curl -fsSL https://raw.githubusercontent.com/pxuanbach/the-skills/main/install.sh | bash

Non-interactive / CLI flags (optional):

# Install specific skills to specific agents
./install.sh --targets claude,antigravity --skills wiki-manager,constructor

# Install all skills to all agents
./install.sh --targets all --skills all

---

Target Agent Directories

Agent	Default Path
Pi Agent	~/.pi/agent/skills
Claude Code	~/.claude/skills
Antigravity / Gemini CLI	~/.gemini/config/skills
OpenAI Codex CLI	~/.codex/skills
OpenCode	~/.config/opencode/skills

---

License

MIT — see LICENSE.