Skip to content

Integralist/agent-skills

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

148 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Agent Skills Configuration

This repository contains global agent configuration — skills, agents, rules, and project instructions designed to be used across all projects when working with AI coding assistants.

It serves two targets:

  • Claude Code — reads .claude/ (CLAUDE.md, agents, rules, and a skills symlink)
  • Generic Agent Skills (e.g. Swival) — reads .agents/ (skills, AGENTS.md)

Skills have a single source of truth: .agents/skills/. .claude/skills is a symlink to .agents/skills, so each skill is edited once and both harnesses see it. Skill content is harness-agnostic — generic language ("prompt the user", "spawn a subagent"). Any Claude-only guidance (e.g. agent teams) lives in a clearly-labeled optional "Agent teams (if your harness supports it)" section so other harnesses can ignore it.

Install

# Generic agents only (e.g. Swival): copies .agents/ to ~/.agents/
make install-agents

# Claude Code (also installs agents as a prerequisite): copies CLAUDE.md, rules,
# agents, and scripts to ~/.claude/, then symlinks ~/.claude/skills →
# ~/.agents/skills
make install-claude

# Everything (install-claude already pulls in install-agents)
make install

Structure

.claude/                            # Claude Code
├── CLAUDE.md                       # @-import pointer to ~/.agents/AGENTS.md
├── agents/
│   └── code-improvement-reviewer.md
├── rules/
│   ├── go.md                       # Go conventions (auto-loaded for *.go)
│   └── markdown.md                 # Markdown conventions (auto-loaded for *.md)
├── scripts/                        # Statusline + session-cost helpers
└── skills -> ../.agents/skills     # Symlink — single source of truth

.agents/                            # Canonical skills + conventions
├── AGENTS.md                       # Shared conventions
└── skills/
    ├── agents-md/
    ├── caveman/
    ├── changelog/
    ├── cleanup/
    ├── code-review/
    ├── commit/
    ├── consensus/
    ├── critique/
    ├── decide/
    ├── delegate/
    ├── go-api/
    ├── go-conventions/             # rule → skill (generic harnesses)
    ├── go-testing/
    ├── grepai/
    ├── grill-me/
    ├── grill-with-docs/
    ├── handoff/
    ├── markdown-conventions/       # rule → skill (generic harnesses)
    ├── markdown-to-skill/
    ├── mysql-index-audit/
    ├── next-task/
    ├── redesign/
    ├── refactor/
    ├── research/
    ├── research-plan/
    ├── slack-search/
    ├── summarize-for-product/
    ├── systematic-debugging/
    ├── teach/
    ├── tech-docs/
    ├── test-feedback/

Components

Skills

Skills are custom code capabilities that extend an agent's ability to perform specific tasks, follow particular patterns, or apply specialized knowledge. Skills act as reusable instructions invoked with /skill-name.

Agents (.claude/agents/)

Custom agents are specialized sub-agents that Claude can spawn via the Task tool. Each agent has a specific purpose and can be configured with different models and instructions.

Current agents:

  • code-improvement-reviewer — Reviews code for readability, performance, and best practices with concrete before/after suggestions

Rules (.claude/rules/)

Rules are modular, topic-specific instruction files that Claude loads automatically. Unlike skills (which are invoked explicitly), rules apply passively — scoped to file patterns via YAML frontmatter paths globs.

In .agents/, rules are converted to skills (go-conventions, markdown-conventions) since generic agents don't support path-scoped auto-loading.

Project Instructions

  • .agents/AGENTS.md — the canonical global conventions
  • .claude/CLAUDE.md — a one-line @~/.agents/AGENTS.md pointer so Claude Code loads the same conventions

Skill Reference

Skill Description
agents-md Make AGENTS.md canonical; stub CLAUDE.md/GEMINI.md as @-import pointers
caveman Ultra-compressed caveman-speak mode; ~75% token reduction
changelog Add a Keep a Changelog entry from the working diff or branch-vs-main
cleanup Audit codebase for AI slop via background subagent
code-review Multi-dimensional review via parallel subagents
commit Git commits with intelligent file grouping
consensus Cross-model second-opinion workflow with discussion rounds and user gates
critique Critique a document for logical fallacies
decide Decision memo with structurer, contrarian, and synthesizer passes
delegate Spawn a subagent for a task
go-api Generate a production-ready Go API service
go-conventions Go coding conventions (mirrors .claude/rules/go.md)
go-testing Write Go tests — table-driven, fuzz, benchmarks
grepai Semantic code search by intent
grill-me Interview the user relentlessly to stress-test a plan or design
grill-with-docs Stress-test a plan against project domain model and update CONTEXT/ADRs
handoff Compact the current conversation into a handoff document for another agent
markdown-conventions Markdown formatting conventions (mirrors .claude/rules/markdown.md)
markdown-to-skill Bulk-convert Markdown files into agent skills
mysql-index-audit Statically audit a codebase for MySQL index misuse (leftmost-prefix, gaps, killers)
next-task Continue working through a project plan
redesign Codebase-wide aspirational audit; produces phased redesign plan with mandatory test pinning
refactor Analyze a feature and produce a reimplementation plan
research Research a topic or repo deeply; writes a reference doc to docs/research/
research-plan Research topics deeply, then create implementation plans (delegates Phase 1 to research)
slack-search Drive Slack via Playwright MCP to run searches or Slackbot prompts
summarize-for-product Translate a plan doc or branch diff into a non-engineer summary (PR/Slack/email)
systematic-debugging Four-phase debugging with root cause analysis
teach Stateful tutor workspace — missions, lessons, learning records, reference docs
tech-docs Write or improve technical documentation via five documentation pillars
test-feedback Parse test failures and fix them in a background subagent

How .claude/ and .agents/ relate

Skills are no longer duplicated. .agents/skills/ is the single source; the .claude/skills symlink points at it. The only things unique to .claude/ are Claude-specific assets that have no generic equivalent:

Asset Purpose
CLAUDE.md One-line @~/.agents/AGENTS.md import — same conventions
rules/ Path-glob auto-loaded conventions (generic harnesses use skills)
agents/ Custom sub-agent definitions spawned via the Task tool
scripts/ Statusline and session-cost helpers

Inside skill bodies, harness-agnostic language is the default. Optional Claude-only behavior (agent teams, allowed-tools/argument-hint frontmatter) is additive — other harnesses ignore what they don't understand.

Claude-specific frontmatter

Skill bodies are harness-agnostic, but some YAML frontmatter keys are read only by Claude Code. They're safe to leave in shared skills — other harnesses ignore unknown keys.

Field Where Purpose
user-invocable SKILL.md Exposes the skill as a /skill-name slash command
argument-hint SKILL.md Placeholder text shown after the slash command in the prompt
allowed-tools SKILL.md Pre-approves specific tool calls (e.g. Bash(git diff:*))
disable-model-invocation SKILL.md Prevents the model from auto-invoking; user must call it explicitly
arguments SKILL.md Structured argument definitions for a slash command
paths rules/*.md Glob patterns that auto-load a rule when matching files are touched

Generating rules from skills

.claude/rules/go.md and .claude/rules/markdown.md are generated from go-conventions and markdown-conventions skills — the skill SKILL.md is the single source of truth, and the rule differs only by frontmatter (paths: globs in place of name:/description:). Bodies stay byte-identical.

Regenerate with make rules (runs .claude/scripts/gen-rules.sh). make install runs it automatically. After editing a *-conventions skill, run make rules before committing, since the generated rules are committed.

Workflow

  • research-plan → critique → next-task → commit → code-review → cleanup → refactor → redesign

Contributing

When adding to this repository:

  1. Ensure additions are truly global and applicable across multiple projects
  2. Write clear, concise descriptions to ensure accurate interpretation
  3. Include examples where helpful
  4. Avoid project-specific details or configurations
  5. Add new skills under .agents/skills/<name>/ only — the .claude/skills symlink picks them up automatically. Keep content harness-agnostic; put any Claude-only guidance in an optional "Agent teams (if your harness supports it)" section
  6. Test with Claude to ensure the desired behavior

License

This repository contains personal coding skills and preferences. Feel free to use and adapt these skills for your own projects.

About

Agent Skills

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors