Skip to content

oramasearch/agent-skills

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

31 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Skills CLI

If you need more info, help, or want to provide general feedback on Orama, join the Orama Slack channel

Orama Agent Skills

A curated collection of AI agent skills for Claude Code and compatible agents, distributed by Orama.

Agent Skills are self-contained instruction sets that give AI coding agents specialized capabilities. Each skill defines a complete workflow — from persona construction to execution strategy — that an agent can follow to accomplish complex, multi-step tasks autonomously.

This repository is designed for use with the skills CLI. Install individual skills or the entire collection into your project, and your AI agent gains new abilities instantly.

Skills Catalog

Skill Description
orama-cloud-cli Drive the Orama Cloud CLI (orama) to provision an account, index a CSV or postgres:// source, and run search — fully from an agent context, no dashboard, no human in the loop. Covers the locked --json / --agent contract, the frozen exit-code table, env resolution, search modes, and cold-start auto-signup.
amaro Drive a running Amaro instance from the shell via the amaro headless CLI — inspect/run data apps, manage connectors, drive the live desktop UI, stream telemetry, control lifecycle. Router skill over three transports (cloud REST / local MCP / Tauri-IPC) with eight namespace reference files and a versioned --json output contract.

Getting Started

⚠️ Run the install from your project folder — not your home directory. Both installers below write a .claude/skills/ and .agents/skills/ folder into the current working directory. cd into the dedicated folder where you actually run your coding agent first (e.g. ~/code/my-project). Do not run it from ~, /, or any generic system path — that scatters skill files across your home directory and makes them apply to every session indiscriminately.

mkdir -p ~/code/my-project && cd ~/code/my-project   # a real project folder
# …then run one of the installers below

The curl | sh installer enforces this — it aborts if you run it from $HOME, /, a system directory, or a standard home subfolder (Desktop/Documents/Downloads/…), and suggests a dedicated folder. Override with --force if you really mean it.

Two ways to install — pick either. Both drop the same skill bundles into ./.claude/skills/<name>/ (Claude Code) and ./.agents/skills/<name>/ (Codex / compatible agents) in the current directory.

Option A — curl | sh (no Node)

The no-frills path. No Node, no flags to learn. The root install.sh downloads the repo once and runs each skill's own installer — installing every skill for both agents by default:

curl -fsSL https://raw.githubusercontent.com/oramasearch/agent-skills/main/install.sh | sh

Install only some skills:

curl -fsSL https://raw.githubusercontent.com/oramasearch/agent-skills/main/install.sh | sh -s -- --skills amaro,orama-cloud-cli

Install a single skill by running just that skill's installer (each skill ships its own <skill>/install.sh):

curl -fsSL https://raw.githubusercontent.com/oramasearch/agent-skills/main/amaro/install.sh | sh

Other options: --dir <path> (target a different project root), --ref <branch|tag|sha>, --list (show available skills), --force, --help. Re-running upserts each skill in place.

Option B — npx skills (needs Node ≥ 18)

The skills CLI. Adds a skills-lock.json tracking what's installed (see Verify installation).

Prerequisites:

Install one skill:

npx skills add oramasearch/agent-skills --skill <skill-name>

Install all skills:

npx skills add oramasearch/agent-skills --all

Once installed (either way), skills are available to your AI agent automatically. Invoke them by describing a task that matches the skill's trigger.

Verify installation

After installation, you should see the skill files in your project's skills directory and a skills-lock.json tracking installed skills:

{
  "version": 1,
  "skills": {
    "<skill-name>": {
      "source": "github/oramasearch/agent-skills",
      "sourceType": "github"
    }
  }
}

Skill Anatomy

Each skill follows a consistent structure:

skill-name/                # one directory per skill, at the repo root
  SKILL.md          # Skill definition (required) — YAML frontmatter + instructions
  README.md         # Human-browsable doc for the skill (required)
  references/       # Bundled reference docs loaded on demand (optional)
  scripts/          # Helper scripts (optional)
  assets/           # Templates / images (optional)

The SKILL.md file is the entry point installed by npx skills. Its YAML frontmatter defines the skill's name and description (used for trigger matching), followed by the full instructions the agent will follow. The folder name matches the frontmatter name.

Note

Reference files are loaded by the agent at runtime — they keep the main SKILL.md focused while providing depth on demand.

Creating Skills

Agents adding skills to this repo follow CLAUDE.md — it is the authoritative guide for layout, the npx skills discovery rules, and the per-skill checklist. In short, each skill should:

  1. Solve a specific, repeatable problem — skills work best when they encode a well-defined workflow
  2. Be self-contained — include all instructions, templates, and scripts the agent needs
  3. Use rich frontmatter — write a descriptive description field with trigger phrases so agents know when to activate the skill
  4. Carry a README.md — a human-browsable doc with the install line, what it does, bundled files, and requirements
  5. Include bundled resources — break complex workflows into reference docs rather than stuffing everything into SKILL.md
  6. Be production-oriented — skills should produce real, working output — not drafts or placeholders
---
name: my-skill
description: Short description of what this skill does and when to use it.
---

# My Skill

Instructions the agent follows...

License

Orama Agent Skills is licensed under the Apache 2.0 license.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages