* Fix Windows installation with smart caching installer
Fixes#52 - Windows users getting ERR_MODULE_NOT_FOUND for better-sqlite3
## Problem
Windows users (@adrianveen and others) were experiencing installation failures
with cryptic ERR_MODULE_NOT_FOUND errors. The root cause was:
1. npm install running on EVERY SessionStart (slow, wasteful)
2. Silent logging hiding actual installation errors
3. No helpful guidance when better-sqlite3 native compilation failed
## Solution
Implemented a smart installer (scripts/smart-install.js) that:
- Caches installation state with version marker (.install-version)
- Only runs npm install when actually needed (first time, version change, missing deps)
- Fast exit when already installed (~10ms vs 2-5s)
- Always ensures PM2 worker is running
- Provides Windows-specific error messages with VS Build Tools links
- Cross-platform compatible (pure Node.js)
## Changes
- Added: scripts/smart-install.js - Smart caching installer with PM2 worker management
- Modified: plugin/hooks/hooks.json - Use smart-install.js instead of raw npm install
- Modified: .gitignore - Added .install-version cache file
- Modified: CLAUDE.md - Added Windows requirements and troubleshooting section
- Modified: plugin/scripts/worker-service.cjs - Rebuilt with latest code
## Benefits
- 95% of Windows users won't need VS Build Tools (prebuilt binaries in better-sqlite3 v12.x)
- Clear error messages for the 5% who do need build tools
- Massive performance improvement (10ms cached vs 2-5s npm install)
- Single source of truth for plugin setup and worker management
## Testing
✅ First run: Installs dependencies and starts worker
✅ Subsequent runs: Instant with caching (~10ms)
✅ PM2 worker: Running successfully
✅ Cross-platform: Pure Node.js, no shell scripts
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Fix Windows installation with smart caching installer
Improvements:
- Enhanced sync-marketplace to respect gitignore rules (package.json)
- Added dynamic Python version detection in Windows help text (scripts/smart-install.js)
- Fixed hardcoded Python version message to show actual installed version
Technical changes:
- Modified package.json sync-marketplace script to use --filter=':- .gitignore' --exclude=.git
- Added runtime Python version detection in getWindowsErrorHelp function
- Improved user experience by showing actual Python installation status
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
Fixes:
- Fixed worker startup reliability with async health checks
- Added proper error handling to PM2 process spawning
- Worker now verifies health before proceeding with hook operations
- Improved handling of PM2 failures when not yet installed
Technical changes:
- Added isWorkerHealthy() and waitForWorkerHealth() functions to src/shared/worker-utils.ts
- Changed ensureWorkerRunning() from synchronous to async with proper await
- All hooks now await ensureWorkerRunning for reliable worker communication
- Rebuilt all plugin executables with version 5.0.2
- Updated version to 5.0.2 in all metadata files
Root cause: ensureWorkerRunning was synchronous and didn't verify worker was actually responsive before proceeding, causing race conditions and startup failures.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Updated new-hook.js, save-hook.js, and summary-hook.js to enhance logging and session handling.
- Simplified session auto-creation logic in worker-service.ts to prioritize observation data preservation.
- Added a blocking wait in ensureWorkerRunning function to prevent race conditions during worker startup.
- Improved error handling and logging consistency across hooks.
- Document intentional watch mode in ecosystem config for auto-restart on plugin updates.
- Implement auto-session creation in worker service to preserve observation data, ensuring no loss of valuable information during hook failures.
- Address race condition concerns in worker startup and improve error handling in hooks.
- Refactored logging functionality in summary-hook.js to improve clarity and consistency.
- Added checks in worker-utils.ts to prevent unnecessary restarts of the worker service, ensuring it only starts if not already running.
- Updated `save-hook.js`, `summary-hook.js`, `context-hook.ts`, `new-hook.ts`, and `save-hook.ts` to include a call to `ensureWorkerRunning()` at the beginning of their main functions. This ensures that the worker is active before any operations are performed.
- Cleaned up import statements in the affected files to include the new utility function from `worker-utils.js`.
- Minor adjustments to logging and error handling to improve robustness and clarity.
- Removed ensureWorkerRunning calls from multiple hooks (cleanup, context, new, save, summary) to streamline code and avoid unnecessary checks.
- Introduced fixed port usage for worker communication across hooks.
- Enhanced error handling in newHook, saveHook, and summaryHook to provide clearer messages for worker connection issues.
- Updated worker service to start without health checks, relying on PM2 for management.
- Cached Claude executable path to optimize repeated calls.
- Improved logging for better traceability of worker actions and errors.
Documentation:
- v5-reddit-post.md: v5.0-specific post focusing on hybrid search breakthrough
- v5-reddit-FINAL-DRAFT.md: General claude-mem post with timeline examples
- v5-reddit-post-story.md: Architecture evolution narrative
- v5-reddit-post-draft.md: Early draft with search examples
- v5-linkedin-post.md: Professional LinkedIn announcement
- reddit-posts.md: Research and reference materials
These are working drafts for community announcements.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Changed user prompt formatting to use full text instead of truncated version.
- Updated date filtering logic to use milliseconds instead of seconds for 90-day cutoff.
- Renamed doc_type values in ChromaSync to ensure consistency and prevent deduplication issues.
- Improved documentation for concept tags in input schema.
- Updated the StdioClientTransport configuration in search-server.ts to include 'stderr: ignore' for the Chroma client.
- Modified the ChromaSync class in ChromaSync.ts to also set 'stderr: ignore' when initializing the Chroma client.
- Implemented a new tool to search for observations using natural language and retrieve timeline context around the best match.
- Introduced two modes: "auto" for automatic timeline anchor selection and "interactive" for user selection of top matches.
- Added input schema validation using zod for query parameters including depth before/after, limit, and project filtering.
- Integrated hybrid semantic search with fallback to FTS5 for observation retrieval.
- Enhanced response formatting for both modes, including detailed timeline context and observation summaries.
- Added `getObservationById` method to retrieve observations by ID in SessionStore.
- Introduced `getSessionSummariesByIds` and `getUserPromptsByIds` methods for fetching session summaries and user prompts by IDs.
- Developed `getTimelineAroundTimestamp` and `getTimelineAroundObservation` methods to provide a unified timeline of observations, sessions, and prompts around a specified anchor point.
- Enhanced ChromaSync to format and sync user prompts, including a new `syncUserPrompt` method.
- Updated WorkerService to sync the latest user prompt to Chroma after updating the worker port.
- Created tests for timeline querying and MCP handler logic to ensure functionality.
- Documented the implementation plan for user prompts and timeline context tool in the Chroma search completion plan.
Changes:
- Updated legend to show only observation types (bugfix, feature, refactor, change, discovery, decision)
- Mapped each type to a color dot emoji (🔴 bugfix, 🟢 feature, 🔵 refactor, ⚪ change, 🟡 discovery, 🟤 decision)
- Removed concept-based filtering and icon selection
- Simplified progressive disclosure instructions to reference types instead of concepts
- All observations now shown in timeline (no concept filtering)
Technical details:
- Modified: src/hooks/context-hook.ts:203-207 (legend)
- Modified: src/hooks/context-hook.ts:210-223 (progressive disclosure text)
- Modified: src/hooks/context-hook.ts:344-369 (icon mapping switched from concepts to types)
- Modified: src/hooks/context-hook.ts:168-173 (removed concept filtering)
- Rebuilt: plugin/scripts/context-hook.js
Rationale: Types are mutually exclusive and core to categorization, while concepts are multi-select metadata better accessed through MCP search tools. This simplifies the display and reduces visual noise.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Features:
- New UsageLogger utility that writes usage metrics to daily JSONL files
- Captures token counts, costs, timing, and cache metrics from SDK result messages
- Usage logs stored in ~/.claude-mem/usage-logs/ (one file per day)
- Added analyze-usage.js script for analyzing usage patterns
Usage data captured:
- Token counts (input, output, cache creation, cache read)
- Total cost in USD per API call
- Duration metrics (total and API time)
- Number of turns per session
- Session and project attribution
Analysis script features:
- Aggregates totals by project and model
- Shows cache hit rates and savings
- Displays cost breakdowns and averages
- npm scripts: usage:analyze and usage:today
Files:
- src/utils/usage-logger.ts (new)
- src/services/worker-service.ts (modified - captures SDK result messages)
- scripts/analyze-usage.js (new)
- package.json (added usage:* npm scripts)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixes:
- Fixed SessionStart hooks running on session resume
- Added matcher configuration to only run hooks on startup, clear, or compact events
- Prevents unnecessary hook execution and improves performance
Technical changes:
- Modified plugin/hooks/hooks.json (added matcher)
- Updated version to 4.3.4 in all metadata files
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Built search-server.mjs successfully (55KB)
- Configured with packages: 'external' to use node_modules dependencies
- MCP config points to ${CLAUDE_PLUGIN_ROOT}/scripts/search-server.mjs
- Ready for deployment to plugin directory
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Corrected incorrect documentation that claimed /clear doesn't inject context.
The SessionStart hook fires with source: "clear" when /clear is used, which
re-injects context from recent sessions. Updated docs to accurately reflect
that /clear both clears conversation AND re-injects fresh context.
Co-authored-by: thedotmack <683968+thedotmack@users.noreply.github.com>
- Changed "Stop Claude" to "Claude finishes responding" to clarify that the Stop hook fires automatically
- Explained Stop hook triggers summary generation automatically, not as a user action
- Added detailed section explaining /clear behavior vs completely exiting Claude Code
- Clarified when new context is injected (new session start, not after /clear)
- Fixed session count: changed "last 3 sessions" to "last 10 sessions" (matches implementation)
Co-authored-by: thedotmack <683968+thedotmack@users.noreply.github.com>
Alex Newman