Zone-based syntax for controlling what AI agents see. Better benchmarks, clean handoffs, backwards compatible.
npx ps-lang@alpha init
PS-LANG v0.1.0-alpha.1 is now available. Install with npx ps-lang@alpha init and start controlling agent context today.
<. Current agent only - hidden from next agent .> Research notes, debug info, internal reasoning <#. Pass to next agent - clean context only #.> Processed findings ready for Agent B <@. Active workspace - current agent can edit @.> Collaborative zone for current work <~. AI-managed metadata - auto-generated ~.> Timestamps, tags, benchmarks <$. Business context - monetization strategy $.> Pricing ideas, revenue notes
Pass only relevant context between agents. No contamination, no noise, just clean inputs.
Test each agent's true performance without upstream hints or reasoning contamination.
Define information flow in agent chains. Research → Analysis → Output, with precision.
Commands automatically structure output with PS-LANG context zones. No manual tagging needed.
.login # Start your day .daily # Review schedule .journal # Auto-tagged daily entry .blog # Auto-tagged content generation .commit # Auto-tagged git commit .handoff # Agent-to-agent handoff .logout # End session # Example: .journal auto-generates zones <.journal 09-26-25-ps-lang <#. Pass to writing agent: Built PS-LANG specification #.> <. Hidden from agents: Debug notes, API keys, raw research data .> <@. Agent workspace: Generate blog post from today's work @.> >
Marketing sprint, blog posts, demo playground with use cases
Filter MVP, CLI commands, LangChain adapter
.psl format, Clerk encryption, secrets management