Ensure you have the following installed on your machine:
Install the package directly from PyPI:
pip install agentsync-vcs
Or install from source:
git clone https://github.com/sumanthp/agent-sync.git cd agent-sync pip install -e .
Navigate to your project root and run the initialization command:
$ agentsync-vcs init
This creates a sample-rule.md and a .agent-sync/config.json file.
If your team has a shared repository for agent rules, add it as a remote:
$ agentsync-vcs remote add https://github.com/my-org/agent-rules.git $ agentsync-vcs sync
The sync command clones the repository to a local cache on your machine.
Other Remote Commands:
$ agentsync-vcs remote list # View configured remotes $ agentsync-vcs remote remove https://github.com/my-org/agent-rules.git # Remove a remote
Create Universal Markdown rules with YAML frontmatter. You can define advanced routing via the type field (e.g., skill or global):
--- name: api-expert description: Specialized FastAPI skill type: skill globs: ["src/api/**/*.py"] --- - Use Pydantic for validation. - Return HTTP 400 for bad input.
Use the pull command followed by your target tool. This will compile both local and remote rules into the specific IDE formats:
# For Cursor (.cursor/rules/*.mdc) $ agentsync-vcs pull cursor # For Claude Code (.claude/skills/) $ agentsync-vcs pull claude # For Windsurf (.windsurf/rules/*.md) $ agentsync-vcs pull windsurf # For Trae (.trae/rules/*.md) $ agentsync-vcs pull trae # For Hermes (.hermes/skills/*.md) $ agentsync-vcs pull hermes # For Kiro (.kiro/steering/*.md) $ agentsync-vcs pull kiro # For GitHub Copilot (.github/instructions/*.md) $ agentsync-vcs pull copilot # For Gemini (GEMINI.md) $ agentsync-vcs pull gemini # For Codex (AGENTS.md) $ agentsync-vcs pull codex
.md files with YAML frontmatter to the project root, and agentsync-vcs will automatically include them in the next pull.