Skills
Skills are domain-specific capabilities defined as SKILL.md files in .cursor/skills/. Each skill teaches the agent how to use a particular tool, workflow, or integration. Unlike rules (which inject context passively), skills are loaded on demand when the conversation matches certain keywords or contexts.
Skills differ from sub-agents in that they provide knowledge and patterns to the main agent, while sub-agents are separate processes that run independently with their own context.
How Skills Load
When the agent recognizes a conversation matches a skill's trigger keywords, it reads the SKILL.md file and follows the instructions within. The agent rules (agent-rules.mdc) reference specific skills and when to use them.
Skills by Category
Browser and Automation
| Skill | Triggers |
|---|---|
| Playwright MCP | Testing the Fastlane Portal, UI verification, form filling, carrier portal automation |
| Safeco Full Flow | End-to-end Safeco bind test, DA-to-Fastlane flow |
Data Parsing
| Skill | Triggers |
|---|---|
| Parse DOCX | .doc/.docx files, Word documents |
| Parse XLSX | .xlsx files, Excel spreadsheets |
| Parse XML | XML files, ACORD, SOAP, XPath queries |
Database
| Skill | Triggers |
|---|---|
| Local DB | Database, tables, schema, SQL, queries, records |
Design
| Skill | Triggers |
|---|---|
| Figma MCP | Figma, design implementation, Figma URLs |
DevOps
| Skill | Triggers |
|---|---|
| GitLab CLI | Merge requests, CI pipelines, releases, GitLab operations |
| Atlassian CLI | Jira, issues, sprints, tickets, story points |
File Layout
.cursor/skills/
├── playwright-mcp/SKILL.md
├── safeco-full-flow/SKILL.md
├── parse-docx/SKILL.md
├── parse-xlsx/SKILL.md
├── parse-xml/SKILL.md
├── local-db/SKILL.md
├── figma-mcp/SKILL.md
├── glab-cli/SKILL.md
└── acli/SKILL.md