cmsfornerd/cmsfornerd2/warp-agent-skills

Warp Agent Skills Specification (warp-agent-skills)

The warp-agent-skills skill defines the architectural specification for building reusable, parameterised, and shareable instructions that AI agents (Warp Agent, OpenViking, Google Jules, Claude Code) discover and invoke autonomously or via slash commands (/{skill-name}).

Skill Directory Structure & Discovery Order

1. Project-Scoped Skills (Repository Level):

• .agents/skills/{skill-name}/SKILL.md (recommended)
• .warp/skills/{skill-name}/SKILL.md
• .github/skills/{skill-name}/SKILL.md
• Discovered automatically when the current working directory is inside the repository.

1. User-Scoped Skills (Global Machine Level):

• ~/.agents/skills/{skill-name}/SKILL.md
• ~/.warp/skills/{skill-name}/SKILL.md

Each skill resides in its dedicated directory containing SKILL.md and optional supporting automation scripts, templates, or schema files.

Parameterised Skill Arguments Syntax

Skills support argument placeholders that are substituted dynamically when invoked:

• $ARGUMENTS: The full raw argument string passed after the skill command.
• $ARGUMENTS[N] or $N: The Nth whitespace-separated argument (0-indexed). For example, $0 represents the first parameter, $1 the second.

Parameterised Skill Frontmatter Example

---
type: "skill"
title: "Explain Topic Skill"
name: "explain-topic"
description: "Explain a specific technical topic for a given audience in a specified tone."
status: "stable"
spec_version: "0.2"
stale_after: "2027-09-17"
topics: ["explanation", "documentation", "warp-skill"]
---
 
# Explain Topic
 
Explain $0 for an audience of $1 professionals using a $2 tone.
 
Full Request String: $ARGUMENTS

Invoking via /explain-topic WebGPU SRE authoritative evaluates to:

"Explain WebGPU for an audience of SRE professionals using a authoritative tone."

OpenViking Skill Storage & Attested MCP Conversion

OpenViking manages skills using canonical URIs (viking://~/skills/{skill-name} or viking://agent/skills/{skill-name}) and provides automatic conversion of Model Context Protocol (MCP) tool definitions containing inputSchema into structured OKF v0.2 SKILL.md format.

Deep State of Mind (DSOM) For My AI Protocol | Harisfazillah Jamel (LinuxMalaysia) | 2026-09-17 Standard: UK English | DBP-standard Bahasa Melayu Malaysia (Piawai) | GNU General Public License v3.0