Initialize Meaning Semantic Index

Bootstrap the .meaning/ directory for a project with semantic file indexing and intelligent inference.

What This Does

Creates a complete semantic index for a project by:

1. Detecting project type (Python, Node, Rust, docs)
2. Installing appropriate schema and config templates
3. Scanning all non-excluded files
4. Running inference on each file (tags, relationships, intents)
5. Generating initial index.yaml with high-confidence suggestions
6. Flagging low-confidence entries for user review

Workflow

1. Pre-flight Checks

from pathlib import Path
from meaning.meaning_core import meaning_dir_exists, detect_project_type, is_git_repo

project_root = Path.cwd()

# Check if .meaning/ already exists
if meaning_dir_exists(project_root):
    print("❌ .meaning/ already exists!")
    print("Use /meaning-update to sync with filesystem changes")
    exit(1)

# Detect project type
project_type = detect_project_type(project_root)
is_git = is_git_repo(project_root)

print(f"✓ Detected project type: {project_type}")
print(f"✓ Git repository: {is_git}")

2. Initialize Structure

from meaning.meaning_core import initialize_meaning, save_index

# Create .meaning/ with templates
index, schema, config = initialize_meaning(project_root, project_type=project_type)

print(f"✓ Created .meaning/ directory")
print(f"✓ Installed {project_type} schema")
print(f"✓ Installed default config")

3. Scan and Infer

from meaning.meaning_core import scan_project_files
from meaning_inference import infer_file_metadata, infer_timestamps

# Scan project files
all_files = scan_project_files(project_root, config)
print(f"✓ Found {len(all_files)} files to index")

# Run inference on each file (limit for initial run)
limit = 50  # Process first 50 files by default
results = []

for i, file_path in enumerate(all_files[:limit]):
    print(f"  Inferring {i+1}/{min(len(all_files), limit)}: {file_path}")
    result = infer_file_metadata(file_path, project_root, index, schema)
    results.append(result)

if len(all_files) > limit:
    print(f"⚠️  Limited to first {limit} files. Run /meaning-update to index remaining {len(all_files) - limit} files.")

4. Build Index with Suggestions

from meaning.meaning_core import FileEntry

now = infer_timestamps()

for result in results:
    # Determine if entry needs review
    needs_review = False
    
    # High-confidence tags
    high_conf_tags = [t.tag for t in result.tags if t.confidence >= 0.8]
    
    # High-confidence relationships
    high_conf_rels = [r.relationship for r in result.relationships if r.confidence >= 0.8]
    
    # Intent (may need review if confidence < 0.8)
    intent = result.intent.intent if result.intent and result.intent.confidence >= 0.7 else ""
    if not intent or (result.intent and result.intent.confidence < 0.8):
        needs_review = True
    
    # Create entry
    entry = FileEntry(
        path=result.path,
        intent=intent or f"[REVIEW NEEDED] {result.path}",
        tags=high_conf_tags if high_conf_tags else ["x-needs-tags"],
        status="active",
        needs_review=needs_review,
        last_verified=now,
        relationships=high_conf_rels,
    )
    
    index.add_file(entry)

print(f"✓ Created {len(index.files)} file entries")

5. Save and Report

from meaning.meaning_core import save_index, validate_index

# Update timestamps
index.last_updated = now

# Save index
save_index(index, project_root)

print(f"✓ Saved index.yaml")

# Validate
validation = validate_index(index, schema, config, project_root)

print("\n" + "="*60)
print("📋 INITIALIZATION COMPLETE")
print("="*60)
print(f"✓ Files indexed: {len(index.files)}")
print(f"✓ Concepts: {len(index.concepts)} (none yet - add manually or use /meaning-review)")
print(f"⚠️  Files needing review: {len(index.files_needing_review())}")
print(f"✓ Validation: {validation.is_valid}")

if validation.warnings:
    print(f"\n⚠️  Warnings: {len(validation.warnings)}")
    for w in validation.warnings[:5]:
        print(f"   • {w}")
    if len(validation.warnings) > 5:
        print(f"   ... and {len(validation.warnings) - 5} more")

print("\n" + "="*60)
print("Next steps:")
print("1. Review entries with needs_review: true")
print("2. Run /meaning-review to accept/modify suggestions")
print("3. Add concept groupings manually to index.yaml")
print("4. Commit .meaning/ to git")
print("="*60)

Arguments

Parse user arguments if provided:

• --type TYPE - Force project type (python, node, rust, docs)
• --limit N - Process first N files (default: 50)
• --all - Process all files (no limit)
• --dry-run - Show what would be created without writing

Error Handling

Common errors and solutions:

Error	Solution
.meaning/ exists	Use /meaning-update instead
No write permissions	Check directory permissions
Unknown project type	Specify --type explicitly
No files found	Check config exclusion patterns

Example Output

✓ Detected project type: python
✓ Git repository: true
✓ Created .meaning/ directory
✓ Installed python schema
✓ Installed default config
✓ Found 127 files to index
  Inferring 1/50: src/main.py
  Inferring 2/50: src/api/client.py
  ...
  Inferring 50/50: tests/test_utils.py
⚠️  Limited to first 50 files. Run /meaning-update to index remaining 77 files.
✓ Created 50 file entries
✓ Saved index.yaml

============================================================
📋 INITIALIZATION COMPLETE
============================================================
✓ Files indexed: 50
✓ Concepts: 0 (none yet - add manually or use /meaning-review)
⚠️  Files needing review: 12
✓ Validation: True

⚠️  Warnings: 77
   • File not indexed: docs/api.md
   • File not indexed: scripts/deploy.sh
   ... and 75 more

============================================================
Next steps:
1. Review entries with needs_review: true
2. Run /meaning-review to accept/modify suggestions
3. Add concept groupings manually to index.yaml
4. Commit .meaning/ to git
============================================================

Notes

• Inference is not perfect - Review suggestions before committing
• Start small - Default 50-file limit prevents overwhelming output
• Iterative - Run /meaning-update to index remaining files after review
• Git-friendly - All YAML files are human-readable and diffable
• No external dependencies - Works offline, no API calls

Philosophy

Do not write code before stating assumptions.
Do not claim correctness you haven't verified.
Do not handle only the happy path.
Under what conditions does this work?

This skill:

• ✓ States assumptions (project type, file accessibility)
• ✓ Verifies correctness (validation after creation)
• ✓ Handles errors (checks for existing .meaning/, permission issues)
• ✓ Documents conditions (requires write access, valid project structure)