dcurlewis/.claude/skills/import

Scan an external folder to extract relevant context (people, projects, terminology) and optionally copy documents for processing. The source folder is treated as read-only; no files are modified or deleted.

Process

Step 1: Get Source Path

Ask the user for the path to their existing knowledge base:

Please provide the path to the folder you'd like to import from.
This is a read-only scan — your original files will not be modified.

Path: 

Verify the path exists and list a summary of its contents (file count by extension, total size, top-level structure).

Step 2: Scan and Classify

Scan markdown files (.md) in the source folder. For each file, extract:

1. People mentioned — proper nouns that appear to be names, especially in headings or structured content
2. Projects or initiatives — recurring multi-word terms that appear to be project names
3. Teams or groups — organisational unit names
4. Acronyms and terminology — abbreviations used consistently
5. Key relationships — who works with whom, reporting lines, collaboration patterns

Read up to 50 files (prioritise recent files by modification date). For each file, read the first 200 lines to extract context.

Step 3: Present Findings

Present a summary of what was found:

Scanned N files from [path]. Here's what I found:

People (N):
- [Name 1] — [context where mentioned]
- [Name 2] — [context where mentioned]

Projects/Initiatives (N):
- [Project Name] — [brief description]

Teams/Groups (N):
- [Team Name] — [context]

Terminology (N):
- [ACRONYM] — [likely meaning]

Would you like me to:
1. Create entity stubs for the people and projects found
2. Add terminology to reference.md
3. Copy specific documents to Raw-Materials/Docs/ for full processing via /doc
4. All of the above
5. Let me review and select specific items first

Step 4: Apply Imports

Based on the user's choice:

Entity stubs: For each confirmed person/project/team:

• Create a minimal stub in the appropriate Curated-Context/Entities/ subdirectory
• Add to the Canonical Names list in Guidelines/wikilink-guidelines.md
• Use information extracted from the source files to populate frontmatter where possible

Terminology: Add confirmed acronyms and terms to Curated-Context/System/reference.md

Document copy: Copy selected files to Raw-Materials/Docs/ for later processing via /doc. Inform the user how many files were copied and that they can run /doc to process them.

Step 5: Summary

Import complete:
- N entity stubs created
- N terms added to reference.md
- N documents copied to Raw-Materials/Docs/

Next steps:
- Run /doc to process imported documents
- Run /setup to continue personalising the system
- Review created stubs in Curated-Context/Entities/ and add any missing details

Constraints

• Read-only source: Never modify, move, or delete files in the source folder
• Confirmation before action: Always show what will be created and get user approval
• No duplication: Check for existing stubs before creating new ones
• Conservative extraction: Only extract entities and terms that appear with reasonable confidence. When in doubt, present to the user for confirmation rather than auto-creating.