Tracking Status (Auto-Captured)
Tracking Enabled: ![ -f .claude/.ref-autotrack ] && echo "✓ Active" || echo "✗ Inactive"
Sources File: ![ -f claude_usage/sources.md ] && echo "✓ Exists ($(wc -l < claude_usage/sources.md) entries)" || echo "✗ Not created"
Prompts File: ![ -f claude_usage/prompts.md ] && echo "✓ Exists ($(grep -c '^Prompt:' claude_usage/prompts.md 2>/dev/null || echo 0) entries)" || echo "✗ Not created"
Git Status: !git rev-parse --is-inside-work-tree &>/dev/null && echo "✓ Git repo" || echo "✗ Not a git repo"
Quick Example
/track:init
Creates: claude_usage/sources.md (with preamble)
claude_usage/prompts.md (with preamble)
.claude/.ref-config (with default settings)
.claude/.ref-autotrack (enables hooks-based tracking)
init - Initialize Hooks-Based Tracking
Initialize automatic reference and prompt tracking for the current project.
What it does
Create claude_usage/ directory:
-
Creates ./claude_usage/ directory for tracked files
-
Visible (not hidden) for easy access and review
Create tracking files with preambles:
-
Creates ./claude_usage/sources.md with explanatory preamble
-
Creates ./claude_usage/prompts.md with explanatory preamble
-
If files exist, leaves them unchanged
-
Preambles explain format, usage, and configuration
Create .claude/ directory structure:
-
Creates ./.claude/ directory if missing
-
Creates ./.claude/.ref-config with default settings: PROMPTS_VERBOSITY=major SOURCES_VERBOSITY=all EXPORT_PATH=exports/
-
Creates ./.claude/.ref-autotrack marker file with metadata
Enable hooks-based tracking:
-
Hooks activate automatically when .ref-autotrack exists
-
PostToolUse hook → tracks WebSearch/WebFetch/Read/Grep
-
UserPromptSubmit hook → captures user prompts
-
SessionEnd hook → pairs prompts with outcomes
-
No skill activation needed - fully automatic
Migrate existing files (if present):
-
Detects old CLAUDE_SOURCES.md and CLAUDE_PROMPTS.md
-
Offers to migrate content to new claude_usage/ directory
-
Preserves existing tracked data
Show setup summary:
-
Files created/existing
-
Auto-tracking status (enabled by default in v2.0)
-
Config location
-
Next steps
Default configuration
After init, tracking is enabled by default (changed in v2.0.0):
-
Sources: Track all WebSearch/WebFetch operations automatically
-
Prompts: Track major academic/development work automatically
-
Export: Default export path is exports/ directory
Hooks run automatically - no manual intervention needed.
Next steps
After running /track:init :
-
Work normally - hooks track automatically
-
Use /track:config to adjust verbosity settings
-
Use /track:auto off to temporarily disable tracking
-
Use /track:export to generate bibliographies or methodology sections
-
Use /track:help for detailed documentation
Files created
.claude/ ├── .ref-autotrack # Marker: hooks-based tracking enabled ├── .ref-config # Verbosity and export settings └── .track-tmp/ # Temporary storage for prompt capture claude_usage/ ├── sources.md # Research sources (with preamble) └── prompts.md # Major prompts and outcomes (with preamble)
Migration from v1.x
If you have existing CLAUDE_SOURCES.md or CLAUDE_PROMPTS.md files:
-
Init will detect them and offer migration
-
Content is moved to claude_usage/ directory
-
Old files can be archived or removed
-
No data loss during migration
Output
Concise status report showing what was created, migration status, and current configuration.
Implementation
Initialization functionality is implemented in scripts/init.sh :
Get skill directory
SKILL_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
Execute init script
bash "$SKILL_DIR/scripts/init.sh"
Script: skills/init/scripts/init.sh (134 lines)
Features:
-
Directory creation (claude_usage/, .claude/, .track-tmp/)
-
File creation with comprehensive preambles
-
Configuration file with defaults
-
Autotrack marker with metadata and timestamp
-
Legacy file detection and migration guidance
-
Success summary with next steps
See scripts/init.sh for full implementation details.