Cross-Platform Interop — Research Agora

AgentSkills.io Standard

Research Agora skills follow the AgentSkills.io standard — an open format that enables skill interoperability across different AI coding platforms. The standard defines:

SKILL.md format: YAML frontmatter with metadata + markdown instruction content
Common metadata fields: name, version, description, author, tags, platforms
Portability guidelines: How to structure prompts for cross-platform compatibility

This standardization allows skills developed for one platform (like Claude Code) to be converted for others (Cursor, Gemini CLI, etc.), though platform-specific features (tool calls, model routing) require adaptation.

Platform Quick-Start Guides

Claude Code (Native)

Research Agora skills are built natively for Claude Code and require no conversion:

/plugin marketplace add rpatrik96/research-agora

/plugin install academic@research-agora

Note: this per-plugin syntax installs a single plugin. Use /plugin marketplace add rpatrik96/research-agora above to install all plugins at once.

Then use skills with slash commands like /paper-introduction or /paper-verify-experiments, or agents like /claim-auditor.

Cursor

Convert skills to Cursor's .mdc rules format:

                    # Convert a single skill

python scripts/convert-skill.py --format cursor --skill paper-references --output /path/to/project

                    # Convert all skills in a plugin

python scripts/convert-skill.py --format cursor --plugin academic --output /path/to/project

Converted files are placed in .cursor/rules/. Cursor will automatically load them as context rules.

Manual conversion: Copy skill content to .cursor/rules/skill-name.mdc with frontmatter:

                    ---

description: Brief description

globs:

alwaysApply: false

---

# Skill content...

Gemini CLI

Convert skills to Gemini CLI's agent format:

                    # Convert a single skill

python scripts/convert-skill.py --format gemini --skill paper-references --output /path/to/project

                    # Convert all skills in a plugin

python scripts/convert-skill.py --format gemini --plugin academic --output /path/to/project

Converted files are placed in .gemini/agents/ as markdown files.

Manual conversion: Copy skill content to .gemini/agents/skill-name.md:

                    # skill-name

Brief description

# Skill content...

GitHub Copilot

Convert skills to GitHub Copilot's instructions format:

                    # Convert all skills to a single file

python scripts/convert-skill.py --format copilot --all --output /path/to/project

All skills are concatenated into .github/copilot-instructions.md. GitHub Copilot reads this file automatically.

Manual conversion: Add skills to .github/copilot-instructions.md:

                    ## skill-name

Brief description

# Skill content...

AgentSkills.io Standard Format

Export skills in the portable AgentSkills.io format:

                    # Convert a single skill

python scripts/convert-skill.py --format agentskills --skill paper-references --output /path/to/export

                    # Convert all skills

python scripts/convert-skill.py --format agentskills --all --output /path/to/export

Generates SKILL.md files with standardized metadata that any AgentSkills.io-compatible tool can consume.

Using the Converter Tool

The scripts/convert-skill.py utility handles conversion to all supported formats:

Command-Line Options

--format {cursor,gemini,copilot,agentskills} — Target format (required)
--skill SKILL_NAME — Convert a single skill by name
--plugin PLUGIN_NAME — Convert all skills in a plugin (academic, development, editorial, formatting, office, research-agents)
--all — Convert all skills across all plugins
--output PATH — Output directory (default: current directory)

Example Workflows

                # Use paper verification skills in Cursor

python scripts/convert-skill.py --format cursor --skill paper-references --output ~/my-project

python scripts/convert-skill.py --format cursor --skill paper-verify-experiments --output ~/my-project

                # Use all development skills in Gemini CLI

python scripts/convert-skill.py --format gemini --plugin development --output ~/my-project

                # Export everything for sharing

python scripts/convert-skill.py --format agentskills --all --output ~/research-agora-skills

Manual Conversion Guide

You can manually adapt any skill for your preferred LLM tool. The core content is plain markdown — any LLM can use it.

Basic Steps

Find the skill source: Browse github.com/rpatrik96/research-agora or click "Source" on any skill card
Extract the content: Copy the markdown body (everything after the YAML frontmatter)
Adapt the frontmatter: Adjust metadata to match your tool's format (see examples above)
Place in the right location: Follow your tool's conventions (.cursor/rules/, .gemini/agents/, .github/copilot-instructions.md, etc.)

Handling Tool-Specific Features

Some skills reference Claude Code-specific features that may need adaptation:

Tool calls: Skills using MCP tools (e.g., mcp__arxiv__search_papers) may need equivalent APIs or manual steps
Model routing: Model tiers (opus/sonnet/haiku) map differently across platforms — adjust as needed
Agent architecture: Multi-agent skills (orchestrator/helper patterns) are Claude Code-specific; adapt to your tool's capabilities

Contributing Cross-Platform Skills

When creating new skills for Research Agora, consider cross-platform compatibility:

Use standard markdown: Avoid platform-specific syntax where possible
Document dependencies clearly: Specify CLI tools, APIs, or libraries required
Avoid platform-specific tool names: Describe what to do, not which tool to call
Test with the converter: Verify skills convert cleanly before submitting

See CONTRIBUTING.md for detailed guidelines.