Attio MCP Server

# Generated Skill Validation Checklist Use this checklist to verify generated skills before packaging and distribution. --- ## Structure Validation ### Required Files - [ ] `SKILL.md` exists in root directory - [ ] `resources/` directory exists - [ ] At least one resource file exists ### SKILL.md Frontmatter - [ ] Has valid YAML frontmatter (between `---` delimiters) - [ ] `name` field is present - [ ] `name` is hyphen-case (lowercase, digits, hyphens only) - [ ] `name` is max 64 characters - [ ] `name` starts with lowercase letter - [ ] `description` field is present - [ ] `description` is max 1024 characters - [ ] `description` contains no angle brackets (< or >) --- ## Workspace-Specific Validation ### Object Slugs - [ ] All referenced object slugs exist in the user's workspace - [ ] Common objects: `companies`, `people`, `deals` - [ ] Custom objects: Verify with `records_discover_attributes` ### Attribute Slugs - [ ] All attribute slugs match the workspace schema - [ ] Use API slugs, NOT display names - [ ] Verify with attio-workspace-schema skill or MCP discovery ### List References - [ ] All list IDs are valid UUIDs - [ ] List IDs come from actual workspace (not templates) - [ ] Each list's `parent_object` matches the workflow's target object - [ ] Verify with `get-lists` MCP tool ### Select/Status Options - [ ] All option values use exact titles from workspace - [ ] Option values are case-sensitive - [ ] Status field options verified with `records_get_attribute_options` - [ ] Select field options verified with `records_get_attribute_options` --- ## Content Quality ### Completeness - [ ] All workflow steps have descriptions - [ ] Tool references are accurate - [ ] Examples use realistic data - [ ] Cross-references to other skills are correct ### Accuracy - [ ] No hardcoded values from other workspaces - [ ] No placeholder UUIDs (like `uuid-here`) - [ ] Field types match (numbers, strings, arrays) - [ ] Multi-select fields documented as requiring arrays ### Clarity - [ ] Instructions are clear and actionable - [ ] Error handling guidance is included - [ ] Edge cases are documented - [ ] Prerequisites are stated --- ## Technical Validation ### Data Types - [ ] Number fields: Use numbers (`85`), not strings (`"85"`) - [ ] Boolean fields: Use booleans (`true`), not strings (`"true"`) - [ ] Date fields: Use ISO 8601 format (`"2024-12-14"`) - [ ] Array fields: Multi-select uses arrays even for single values ### UUID Validation - [ ] All UUIDs match format: `xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx` - [ ] No truncated or partial UUIDs - [ ] No placeholder UUIDs in final output ### Read-Only Fields - [ ] Read-only fields are not included in update examples - [ ] Computed fields are documented as read-only - [ ] `created_at`, `updated_at` not in update payloads --- ## Final Checks ### Before Packaging - [ ] Run `python quick_validate.py <skill-path>` - [ ] Review all generated content for accuracy - [ ] Test at least one workflow manually - [ ] Verify skill loads in Claude without errors ### After Packaging - [ ] `.skill` file is valid ZIP archive - [ ] Archive contains all expected files - [ ] Skill name matches directory name - [ ] No sensitive data in packaged files --- ## Common Issues and Fixes | Issue | Cause | Fix | | ---------------------- | ---------------------------------------- | ----------------------------------- | | "Invalid name" | Name contains uppercase or special chars | Use hyphen-case | | "Description too long" | Over 1024 characters | Shorten description | | "Unknown attribute" | Wrong API slug | Check workspace schema | | "Invalid option" | Wrong option title | Use `records_get_attribute_options` | | "List not found" | Invalid list UUID | Use `get-lists` to get real IDs | | "Record not found" | Invalid record UUID | Verify UUID format and existence | --- ## Validation Script Usage ```bash # Quick validation python scripts/quick_validate.py ./my-skill # Full validation before packaging python scripts/quick_validate.py ./my-skill && \ python scripts/package_skill.py ./my-skill ```