pqpo/pragma

English | 简体中文
Build Your Agent Team Once. Use It Everywhere.
Build portable AI agent teams across models and harnesses — combine the right models, tools, context, and workflows for each task. Run them directly through Pragma Desktop or CLI; bring them into Codex, Claude Code, and other tools without changing your workflow; or integrate them into your own AI system with the SDK.
Download
· Quick start
· Document
· Examples
Pragma is a cross-harness platform for turning AI-native working methods into runnable, reusable agent teams. A team can combine Experts, ExpertTeams, Flows, tools, Skills, shared context, memory, permissions, and human checkpoints—not just a prompt.
Unlike a single chat product or Coding Agent, Pragma treats the Agent Team and its working method as the portable unit. Define it as YAML DSL and build it in Desktop, run it from the terminal through the CLI, or embed the same team in your own application with the SDK.
During a Mission, work can move between specialists without losing decisions, artifacts, or accumulated experience. The same team can coordinate different models and harnesses for different steps while keeping context, permissions, handoffs, and execution under one governance model.
> [!IMPORTANT]
> Pragma is currently a preview. The latest Desktop release provides builds for macOS Apple Silicon and Intel. Verify that a package comes from the official release before installing it.
Quick start
Run Pragma Desktop
Download the package for your Mac from GitHub Releases, install it, and then:
1. Open **Settings** and connect a model provider or an installed local runtime.
2. Choose the workspace Pragma is allowed to use.
3. Talk with Pragma to create your own Expert, ExpertTeam, or Flow.
4. Start a task with the result and use it as much as you like. As Missions accumulate, cross-harness memory turns repeated experience into stable knowledge and reusable Skills.
To open a Desktop release, use this order:
1. Try opening **Pragma** normally.
2. If macOS blocks it, open **System Settings → Privacy & Security** and choose **Open Anyway**.
3. Only if it still cannot open, and you have verified that the package came from the official release, clear the app's quarantine attribute:
You do not need to change the global **Allow apps downloaded from** setting.
The Desktop distribution guide explains the release contents, checksums, and installation process.
Build Desktop from source
Requirements: Node.js 22 or later and pnpm 10.12.1.
Development builds use ~/.pragma-development/ by default so unreleased storage migrations cannot
modify data owned by the installed Desktop app. Set PRAGMA_HOME explicitly only when a task
intentionally needs another isolated data root.
Try the SDK and runtime adapters
The repository includes runnable examples rather than abbreviated API fragments:
See the examples guide for Expert sessions, delegation, ExpertTeams, Flows, human review gates, MCP, Skills, plugins, memory, recovery, and portable bundles.
What Pragma makes reusable
AI working methods
Compose Experts, ExpertTeams, Flows, subflows, tools, and human checkpoints. Each step can bind the model, agent harness, permissions, and context that fit the task. The resulting method can be versioned, evaluated, exported, and shared.
Private knowledge
Pragma treats context as a Host-owned ContextStore instead of trapping it inside one chat product. Events from different tasks, harnesses, and models enter a shared Memory Pipeline and become evidence-backed episodic and semantic memory. As that dynamic memory accumulates, background work can automatically start a policy-controlled promotion workflow that turns it into stable knowledge or reusable Skills; authoritative changes are activated only after the required review.