Question 1

What is PS-LANG?

Accepted Answer

PS-LANG is an open source framework with zone-based syntax for controlling what AI agents see in multi-agent workflows. Use it to structure prompts, control context handoffs, and maintain privacy in AI collaborations. Works with any AI model—Claude, GPT, Cursor, Copilot, or custom agents. It provides 7 privacy zones that let you control context flow between agents, reduce token usage by up to 60%, and create cleaner agent handoffs.

Question 2

How does PS-LANG reduce token usage?

Accepted Answer

PS-LANG uses zone-based syntax to mark content with specific privacy levels. When agents process prompts, they only see content within zones they have access to, reducing unnecessary context and cutting token usage by up to 60% in multi-agent workflows.

Question 3

What are PS-LANG zones?

Accepted Answer

PS-LANG has 7 privacy zones: public (<@. .>), internal (<. .>), confidential (<#. #>), agent-specific (<.agent .agent>), bookmark (<.bm .bm>), log (<.log .log>), and system (). Each zone controls who can see the content inside.

Question 4

Is PS-LANG backwards compatible?

Accepted Answer

Yes, PS-LANG is designed to be backwards compatible. You can add zone syntax to existing prompts without breaking them. Agents that don't understand PS-LANG will simply ignore the zone markers and process the full text.

Question 5

How do I install PS-LANG?

Accepted Answer

Install PS-LANG using npm: 'npx ps-lang@alpha init'. This initializes PS-LANG in your project and sets up the necessary configuration for zone-based context control in your AI workflows.

Question 6

What is PS-LANG Journal?

Accepted Answer

PS-LANG Journal is the OSS journaling application built on PS-LANG for tracking AI workflows, benchmarking improvements, and maintaining secure audit trails. It uses end-to-end encryption to store prompt history and workflow analytics. MIT licensed and free forever.

Question 7

Does PS-LANG work with MCP?

Accepted Answer

Yes, PS-LANG integrates seamlessly with Model Context Protocol (MCP) agent chains. You can use zone syntax within MCP workflows to control context between different agents in your pipeline.

Question 8

How accurate are PS-LANG benchmarks?

Accepted Answer

PS-LANG improves benchmark accuracy by controlling exactly what context each agent sees. This eliminates context contamination between test runs and ensures consistent, reproducible benchmark results.

Token Usage Comparison

About This Demo

Understanding Token Metrics

What are these metrics showing?

How does PS-LANG reduce token usage?

Are these real benchmarks?

How do I get started with PS-LANG?

Help shape the future