About agent skills

Agent skills teach your AI agent Veracity-specific patterns for authentication, API calls, and service accounts so you don't repeat yourself in every prompt.

Agent skills teach your AI agent Veracity-specific patterns so you do not need to explain those patterns in every prompt. When you install a skill, your agent follows correct Veracity conventions for authentication, API calls, and service account configuration automatically: from the first interaction.

What a skill is

A skill is a Markdown file you place in your repository at a path your agent reads. It contains instructions, constraints, and examples scoped to a specific task. The agent loads the file when it recognises a relevant task and applies the instructions to its output.

Skills follow the open agentskills.io specification. Any agent that supports project-level instruction files can read them: including GitHub Copilot, Cursor, and Claude Code.

Skills work alongside the MCP server

Skills define how to implement something. The Veracity MCP server provides current documentation for what the implementation should contain. Used together, your agent follows the correct pattern and checks its work against live Veracity documentation.

Available skills

No Veracity agent skills are publicly available yet. Several skills are in active use internally and are being stabilised before public release. When they are ready, they will be published here.

To find out which skills are currently available in preview, contact the relevant Veracity team:

  • Data Platform team — skills for Data Workbench, workspace access, and regulated maritime data
  • Application Network team — skills for authentication, frontend integration, and application scaffolding

See Common AI tasks for worked examples using prompts, and Working safely for guidance on credentials, tenancy, and regulated data.