Port of Context Skills

AI agent skills for Port of Context tools.

Available Skills

portlang

portlang is the environment-first agent framework. Learn how to create field.toml configurations, define boundaries and verifiers, add custom tools, debug with trajectories, and optimize agent reliability.

Install:

npx skills add https://github.com/portofcontext/skills --skill portlang

What you'll learn:

• Creating and configuring field.toml files for agent tasks
• Defining boundaries and verifiers to control agent behavior
• Adding custom tools (shell, Python, or MCP servers)
• Debugging agent failures using trajectories
• Optimizing agent reliability with convergence testing
• Configuring Code Mode for token-efficient operations
• Structuring multi-layer verification patterns
• Analyzing agent behavior across multiple runs

Documentation: See portlang/SKILL.md for complete skill guide.

Reference materials:

• Verifier Patterns - 15+ real-world examples
• Custom Tools Guide - Shell, Python, and MCP tools
• Trajectory Analysis - Advanced debugging techniques
• Field Recipes - Complete field.toml examples

Helper scripts:

• scripts/new_field.sh - Interactive field template generator
• scripts/validate_field.sh - Enhanced field validation
• scripts/analyze_trajectories.py - Multi-trajectory analyzer

About

portlang is an environment-first agent framework that treats agent behavior as search through a conditioned space. Unlike traditional agent frameworks that manage loops, portlang manages environments.

Core principle: Define the search space. The agent finds the path.

Learn more: https://github.com/portofcontext/portlang