Standalone vs Connected Mode
Terso works in two modes. Standalone is free, offline, and gets you the AGENTS.md compiler immediately. Connected adds richer project memory backed by Omnus.
Standalone mode (recommended start)
No Omnus account, no network, no auth. Free, open source.
npm install -g terso-cli
cd ~/projects/my-app
terso init
# edit AGENTS.md
terso emit
What works in standalone:
terso emit— compileAGENTS.mdtoCLAUDE.md,.cursorrules,.github/copilot-instructions.mdemit --watch— live re-compile on saveemit --check— CI gate to prevent driftterso mcp— MCP server exposesAGENTS.mdto any agent via MCPterso compile— combine.terso/-tracked source docs into a single context artifact
What's missing without Omnus:
- No captured decisions from the terminal
- No cross-project search
- No shared operational knowledge across projects
- No nightly state consolidation
- No portfolio health monitoring
Connected mode
Add an Omnus account to unlock the capture/sync/search pipeline.
terso auth set <your-api-token>
terso capture "switched auth to Supabase — RLS was the dealbreaker"
terso sync
What's added:
- Capture from CLI, web, email, or webhooks
- LLM classification and dedup
- Cross-project knowledge search
- Shared ops compiled across all your projects
- Nightly state consolidation
- Review queue for uncertain items
- Portfolio health monitoring
Switching modes
You don't have to choose. Connected mode is purely additive — emit keeps working regardless.
Check connection status:
terso doctor
Clear the API token to fall back to standalone:
terso auth clear