Wherobots Agent Skills

Agent skills for the Wherobots spatial analytics platform. These skills provide AI coding assistants with the meta-knowledge they need to work effectively with Wherobots tools.

Skills

Skill	Description
wherobots-usage	Decision matrix for choosing between MCP, CLI, SDK, and Dashboard; auth setup; scheduling guidance
wherobots-explore	MCP workflow patterns for data discovery, schema exploration, and spatial query generation
wherobots-develop	CLI discovery patterns, non-obvious flags, job submission workflow, Python and TypeScript SDK usage

Installation

Via skills.sh (Claude Code, any agent)

npx skills add wherobots/agent-skills@wherobots-usage
npx skills add wherobots/agent-skills@wherobots-explore
npx skills add wherobots/agent-skills@wherobots-develop

Via Wherobots VS Code Extension

Skills are bundled in the Wherobots VS Code extension and available automatically in Copilot Chat and Claude Code when the extension is installed.

Design Philosophy

Every line in a skill must pass this test: "Does this make the agent measurably more effective, or can the agent already discover this from --help / MCP tool descriptions?"

Skills contain meta-knowledge only:

• Interface selection guidance (when to use MCP vs CLI vs SDK)
• Non-obvious workflows and sequences
• Gotchas that waste agent turns when discovered by trial and error
• Decision points between equally valid approaches

Skills do NOT contain:

• Tool parameter documentation (already in MCP/CLI --help)
• API reference material (discoverable at runtime)
• Tutorials or step-by-step guides

Benchmarks

The benchmarks/ directory contains tasks for evaluating skill effectiveness. See benchmarks/README.md for details.

Contributing

1. Fork this repository
2. Create a branch (feat/improve-explore-skill)
3. Edit the relevant SKILL.md file
4. Ensure frontmatter is valid: name must match the parent directory, description is required
5. Open a pull request -- CI will validate frontmatter and lint markdown

License

MIT